├── .Rbuildignore ├── .github └── workflows │ └── check-standard.yml ├── .gitignore ├── .travis.yml ├── CONTRIBUTING.md ├── DESCRIPTION ├── LICENSE ├── NAMESPACE ├── NEWS.md ├── R ├── calculate_responses.R ├── ceteris_paribus.R ├── explain.R ├── model_performance.R ├── pbcTest.R ├── pbcTrain.R ├── plot_ceteris_paribus.R ├── plot_explainer.R ├── plot_model_performance.R ├── plot_prediction_breakdown.R ├── plot_variable_response.R ├── prediction_breakdown.R ├── print_ceteris_paribus.R ├── print_explainer.R ├── print_model_performance.R ├── print_prediction_breakdown.R ├── print_variable_response.R ├── theme_mi2.R ├── variable_response.R └── welcome.R ├── README.md ├── _pkgdown.yml ├── codecov.yml ├── data ├── pbcTest.rda └── pbcTrain.rda ├── docs ├── CONTRIBUTING.html ├── MI2logo.jpg ├── articles │ ├── Custom_predict_for_survival_models.html │ ├── Global_explanations.html │ ├── Global_explanations_files │ │ └── figure-html │ │ │ ├── unnamed-chunk-3-1.png │ │ │ ├── unnamed-chunk-5-1.png │ │ │ └── unnamed-chunk-6-1.png │ ├── How_to_compare_models_with_survxai.html │ ├── How_to_compare_models_with_survxai_files │ │ └── figure-html │ │ │ ├── unnamed-chunk-10-1.png │ │ │ ├── unnamed-chunk-11-1.png │ │ │ ├── unnamed-chunk-12-1.png │ │ │ ├── unnamed-chunk-13-1.png │ │ │ ├── unnamed-chunk-14-1.png │ │ │ ├── unnamed-chunk-14-2.png │ │ │ ├── unnamed-chunk-14-3.png │ │ │ ├── unnamed-chunk-15-1.png │ │ │ ├── unnamed-chunk-16-1.png │ │ │ ├── unnamed-chunk-16-2.png │ │ │ ├── unnamed-chunk-16-3.png │ │ │ ├── unnamed-chunk-17-1.png │ │ │ ├── unnamed-chunk-17-2.png │ │ │ ├── unnamed-chunk-17-3.png │ │ │ ├── unnamed-chunk-18-1.png │ │ │ ├── unnamed-chunk-18-2.png │ │ │ ├── unnamed-chunk-18-3.png │ │ │ ├── unnamed-chunk-19-1.png │ │ │ ├── unnamed-chunk-19-2.png │ │ │ ├── unnamed-chunk-19-3.png │ │ │ ├── unnamed-chunk-20-1.png │ │ │ ├── unnamed-chunk-20-2.png │ │ │ ├── unnamed-chunk-20-3.png │ │ │ ├── unnamed-chunk-21-1.png │ │ │ ├── unnamed-chunk-21-2.png │ │ │ ├── unnamed-chunk-21-3.png │ │ │ ├── unnamed-chunk-23-1.png │ │ │ ├── unnamed-chunk-23-2.png │ │ │ ├── unnamed-chunk-23-3.png │ │ │ ├── unnamed-chunk-25-1.png │ │ │ ├── unnamed-chunk-25-2.png │ │ │ ├── unnamed-chunk-25-3.png │ │ │ ├── unnamed-chunk-26-1.png │ │ │ ├── unnamed-chunk-7-1.png │ │ │ ├── unnamed-chunk-8-1.png │ │ │ └── unnamed-chunk-9-1.png │ ├── Local_explanations.html │ ├── Local_explanations_files │ │ └── figure-html │ │ │ ├── unnamed-chunk-2-1.png │ │ │ ├── unnamed-chunk-3-1.png │ │ │ └── unnamed-chunk-4-1.png │ └── index.html ├── authors.html ├── docsearch.css ├── docsearch.js ├── index.html ├── jquery.sticky-kit.min.js ├── link.svg ├── news │ └── index.html ├── pkgdown.css ├── pkgdown.js ├── pkgdown.yml └── reference │ ├── ceteris_paribus.html │ ├── explain.html │ ├── index.html │ ├── model_performance.html │ ├── pbcTest.html │ ├── pbcTrain.html │ ├── plot.surv_ceteris_paribus_explainer-1.png │ ├── plot.surv_ceteris_paribus_explainer.html │ ├── plot.surv_explainer-1.png │ ├── plot.surv_explainer.html │ ├── plot.surv_model_performance_explainer-1.png │ ├── plot.surv_model_performance_explainer.html │ ├── plot.surv_prediction_breakdown_explainer-1.png │ ├── plot.surv_prediction_breakdown_explainer.html │ ├── plot.surv_variable_response_explainer-1.png │ ├── plot.surv_variable_response_explainer.html │ ├── prediction_breakdown.html │ ├── print.surv_ceteris_paribus_explainer.html │ ├── print.surv_explainer.html │ ├── print.surv_model_performance_explainer.html │ ├── print.surv_prediction_breakdown_explainer.html │ ├── print.surv_variable_response_explainer.html │ ├── theme_mi2.html │ └── variable_response.html ├── inst └── CITATION ├── man ├── ceteris_paribus.Rd ├── explain.Rd ├── model_performance.Rd ├── pbcTest.Rd ├── pbcTrain.Rd ├── plot.surv_ceteris_paribus_explainer.Rd ├── plot.surv_explainer.Rd ├── plot.surv_model_performance_explainer.Rd ├── plot.surv_prediction_breakdown_explainer.Rd ├── plot.surv_variable_response_explainer.Rd ├── prediction_breakdown.Rd ├── print.surv_ceteris_paribus_explainer.Rd ├── print.surv_explainer.Rd ├── print.surv_model_performance_explainer.Rd ├── print.surv_prediction_breakdown_explainer.Rd ├── print.surv_variable_response_explainer.Rd ├── theme_mi2.Rd └── variable_response.Rd ├── materials ├── survxai-cheatsheet.pdf └── survxai-cheatsheet.png ├── misc ├── img │ ├── breakdown.png │ ├── ceteris_paribus.png │ ├── model_performance.png │ └── variable_response.png ├── paper.bib └── paper.md ├── survxai.Rproj ├── tests ├── testthat.R └── testthat │ ├── objects_for_tests.R │ ├── test_explainer.R │ ├── test_plot_ceteris_paribus.R │ ├── test_plot_explainer.R │ ├── test_plot_model_performance.R │ ├── test_plot_prediction_breakdown.R │ ├── test_plot_variable_response.R │ ├── test_prints.R │ ├── test_surv_ceteris_paribus.R │ ├── test_surv_model_performance.R │ ├── test_surv_prediction_breakdown.R │ └── test_surv_variable_response.R └── vignettes ├── Custom_predict_for_survival_models.Rmd ├── Global_explanations.Rmd ├── Global_explanations.html ├── How_to_compare_models_with_survxai.Rmd ├── How_to_compare_models_with_survxai.html ├── How_to_compare_models_with_survxai_files └── figure-html │ ├── unnamed-chunk-12-1.png │ ├── unnamed-chunk-13-1.png │ ├── unnamed-chunk-15-1.png │ ├── unnamed-chunk-16-1.png │ └── unnamed-chunk-9-1.png ├── Local_explanations.Rmd └── Local_explanations.html /.Rbuildignore: -------------------------------------------------------------------------------- 1 | ^.*\.Rproj$ 2 | ^\.Rproj\.user$ 3 | ^codecov\.yml$ 4 | .travis.yml 5 | _pkgdown.yaml 6 | docs/ 7 | materials/ 8 | misc/ 9 | LICENSE 10 | ^_pkgdown\.yml$ 11 | ^docs$ 12 | CONTRIBUTING.md 13 | .github 14 | -------------------------------------------------------------------------------- /.github/workflows/check-standard.yml: -------------------------------------------------------------------------------- 1 | # For help debugging build failures open an issue on the RStudio community with the 'github-actions' tag. 2 | # https://community.rstudio.com/new-topic?category=Package%20development&tags=github-actions 3 | on: 4 | push: 5 | branches: 6 | - master 7 | pull_request: 8 | branches: 9 | - master 10 | 11 | name: R-CMD-check 12 | 13 | jobs: 14 | R-CMD-check: 15 | runs-on: ${{ matrix.config.os }} 16 | 17 | name: ${{ matrix.config.os }} (${{ matrix.config.r }}) 18 | 19 | strategy: 20 | fail-fast: false 21 | matrix: 22 | config: 23 | - {os: windows-latest, r: 'release'} 24 | - {os: macOS-latest, r: 'release'} 25 | - {os: ubuntu-20.04, r: 'release', rspm: "https://packagemanager.rstudio.com/cran/__linux__/focal/latest"} 26 | - {os: ubuntu-20.04, r: 'devel', rspm: "https://packagemanager.rstudio.com/cran/__linux__/focal/latest"} 27 | 28 | env: 29 | R_REMOTES_NO_ERRORS_FROM_WARNINGS: true 30 | RSPM: ${{ matrix.config.rspm }} 31 | 32 | steps: 33 | - uses: actions/checkout@v2 34 | 35 | - uses: r-lib/actions/setup-r@master 36 | with: 37 | r-version: ${{ matrix.config.r }} 38 | 39 | - uses: r-lib/actions/setup-pandoc@master 40 | 41 | - name: Query dependencies 42 | run: | 43 | install.packages('remotes') 44 | saveRDS(remotes::dev_package_deps(dependencies = TRUE), ".github/depends.Rds", version = 2) 45 | writeLines(sprintf("R-%i.%i", getRversion()$major, getRversion()$minor), ".github/R-version") 46 | shell: Rscript {0} 47 | 48 | - name: Cache R packages 49 | if: runner.os != 'Windows' 50 | uses: actions/cache@v1 51 | with: 52 | path: ${{ env.R_LIBS_USER }} 53 | key: ${{ runner.os }}-${{ hashFiles('.github/R-version') }}-1-${{ hashFiles('.github/depends.Rds') }} 54 | restore-keys: ${{ runner.os }}-${{ hashFiles('.github/R-version') }}-1- 55 | 56 | - name: Install system dependencies 57 | if: runner.os == 'Linux' 58 | run: | 59 | while read -r cmd 60 | do 61 | eval sudo $cmd 62 | done < <(Rscript -e 'cat(remotes::system_requirements("ubuntu", "20.04"), sep = "\n")') 63 | - name: Install dependencies 64 | run: | 65 | remotes::install_deps(dependencies = TRUE) 66 | remotes::install_cran("rcmdcheck") 67 | shell: Rscript {0} 68 | 69 | - name: Check 70 | env: 71 | _R_CHECK_CRAN_INCOMING_REMOTE_: false 72 | run: rcmdcheck::rcmdcheck(args = c("--no-manual", "--as-cran"), error_on = "warning", check_dir = "check") 73 | shell: Rscript {0} 74 | 75 | - name: Upload check results 76 | if: failure() 77 | uses: actions/upload-artifact@main 78 | with: 79 | name: ${{ runner.os }}-r${{ matrix.config.r }}-results 80 | path: check 81 | -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- 1 | .Rproj.user 2 | .Rhistory 3 | .RData 4 | .Ruserdata 5 | inst/doc 6 | 7 | -------------------------------------------------------------------------------- /.travis.yml: -------------------------------------------------------------------------------- 1 | language: R 2 | sudo: false 3 | cache: packages 4 | env: 5 | global: 6 | - R_CHECK_ARGS="--no-build-vignettes --no-manual --timings" 7 | 8 | notifications: 9 | email: false 10 | 11 | after_success: 12 | - Rscript -e 'covr::codecov()' -------------------------------------------------------------------------------- /CONTRIBUTING.md: -------------------------------------------------------------------------------- 1 | ## How to contribute to survxai 2 | 3 | #### **Did you find a bug?** 4 | 5 | Please follow these rules when reporting bugs: 6 | 7 | * Install to the latest version of [survxai from GitHub](https://github.com/MI2DataLab/survxai) and check whether the problem still occurs. 8 | 9 | * **Check, if the bug was not already reported** by searching [Issues](https://github.com/MI2DataLab/survxai/issues). 10 | 11 | * If you're unable to find an open issue addressing the problem, [open a new one](https://github.com/MI2DataLab/survxai/issues/new). Be sure to include a **title and clear description**, and a **sample code** demonstrating the problem. 12 | 13 | 14 | #### **Did you fix a bug?** 15 | 16 | * Ensure that changes meet the requirements of [The tidyverse style guide](http://style.tidyverse.org) and relevant tests are added. 17 | 18 | * Open a new GitHub pull request with the solution. 19 | 20 | * Ensure the PR clearly describes the problem and solution. Include the relevant issue number if applicable. 21 | 22 | 23 | #### **Do you intend to add a new feature or change an existing one?** 24 | 25 | * Suggest your change in the [Issues](https://github.com/MI2DataLab/survxai/issues) and start writing code. 26 | 27 | 28 | Thanks! 29 | 30 | MI2 team 31 | -------------------------------------------------------------------------------- /DESCRIPTION: -------------------------------------------------------------------------------- 1 | Package: survxai 2 | Title: Visualization of the Local and Global Survival Model Explanations 3 | Version: 0.2.1 4 | Authors@R: c( 5 | person("Aleksandra", "Grudziaz", email = "aleksandra.grudziaz@gmail.com", role = c("aut", "cre")), 6 | person("Alicja", "Gosiewska", email = "alicjagosiewska@gmail.com", role = c("aut")), 7 | person("Przemyslaw", "Biecek", email = "przemyslaw.biecek@gmail.com", role = c("aut", "ths")) 8 | ) 9 | Description: Survival models may have very different structures. This package contains functions 10 | for creating a unified representation of a `survival` models, which can be further processed by various 11 | survival explainers. Tools implemented in survxai help to understand how input variables are used in 12 | the model and what impact do they have on the final model prediction. Currently, four explanation methods are implemented. 13 | We can divide them into two groups: local and global. 14 | License: GPL 15 | Encoding: UTF-8 16 | LazyData: true 17 | RoxygenNote: 7.1.0 18 | Imports: breakDown, 19 | ggplot2, 20 | pec, 21 | scales, 22 | survival, 23 | survminer 24 | Depends: prodlim 25 | Suggests: CFC, 26 | covr, 27 | DALEX, 28 | knitr, 29 | randomForestSRC, 30 | rmarkdown, 31 | rms, 32 | testthat, 33 | tibble 34 | VignetteBuilder: knitr 35 | URL: https://mi2datalab.github.io/survxai/ 36 | BugReports: https://github.com/MI2DataLab/survxai/issues 37 | -------------------------------------------------------------------------------- /NAMESPACE: -------------------------------------------------------------------------------- 1 | # Generated by roxygen2: do not edit by hand 2 | 3 | S3method(explain,default) 4 | S3method(plot,surv_ceteris_paribus_explainer) 5 | S3method(plot,surv_explainer) 6 | S3method(plot,surv_model_performance_explainer) 7 | S3method(plot,surv_prediction_breakdown_explainer) 8 | S3method(plot,surv_variable_response_explainer) 9 | S3method(predictSurvProb,surv_explainer) 10 | S3method(print,surv_ceteris_paribus_explainer) 11 | S3method(print,surv_explainer) 12 | S3method(print,surv_model_performance_explainer) 13 | S3method(print,surv_prediction_breakdown_explainer) 14 | S3method(print,surv_variable_response_explainer) 15 | export(ceteris_paribus) 16 | export(explain) 17 | export(model_performance) 18 | export(prediction_breakdown) 19 | export(theme_mi2) 20 | export(variable_response) 21 | import(ggplot2) 22 | import(pec) 23 | importFrom(breakDown,broken) 24 | importFrom(prodlim,Hist) 25 | importFrom(scales,seq_gradient_pal) 26 | importFrom(stats,aggregate) 27 | importFrom(stats,as.formula) 28 | importFrom(stats,median) 29 | importFrom(stats,model.frame) 30 | importFrom(stats,na.omit) 31 | importFrom(stats,predict) 32 | importFrom(stats,quantile) 33 | importFrom(stats,weighted.mean) 34 | importFrom(survival,survfit) 35 | importFrom(survminer,ggsurvplot) 36 | importFrom(utils,head) 37 | importFrom(utils,tail) 38 | -------------------------------------------------------------------------------- /NEWS.md: -------------------------------------------------------------------------------- 1 | 2 | survxai 0.2.1 3 | ---------------------------------------------------------------- 4 | * improvment in documentation 5 | * CITATION,md was added 6 | 7 | survxai 0.2.0 8 | ---------------------------------------------------------------- 9 | * CONTRIBUTING.md have been added 10 | * Cheatsheet was created 11 | * Colour palette for `broken_prediction()` and `ceteris_paribus()` was added 12 | * New optional arguments for `plot.broken_prediction()` and `plot.ceteris_paribus()` was added 13 | 14 | survxai 0.1.0 15 | ---------------------------------------------------------------- 16 | * survxai package is now public 17 | * `model_performance()`, `variable_response()`, `broken_prediction()` and `ceteris_paribus()` functions were implemented 18 | * Vignettes for local and global explanations were added 19 | -------------------------------------------------------------------------------- /R/calculate_responses.R: -------------------------------------------------------------------------------- 1 | calculate_responses<- function(vname, times_s, observation_s, model_s, explainer_s, grid_points_s, data_s, predict_function_s) { 2 | if(class(data_s[,vname])=="numeric" || class(data_s[,vname])=="integer"){ 3 | probs <- seq(0, 1, length.out = grid_points_s) 4 | new_x <- quantile(data_s[,vname], probs = probs) 5 | quant_x <- mean(observation_s[1,vname] >= data_s[,vname], na.rm = TRUE) 6 | new_data <- observation_s[rep(1, grid_points_s),] 7 | new_data[,vname] <- new_x 8 | y_hat <- t(predict_function_s(model_s, new_data, times_s)) 9 | 10 | res <- data.frame(y_hat=numeric(), time = numeric(), vname = character(), new_x = numeric(), 11 | x_quant = numeric(), quant = numeric(), relative_quant = numeric(), label = character(), 12 | class = character()) 13 | 14 | for(i in 1:grid_points_s){ 15 | tmp <- data.frame(y_hat = y_hat[,i]) 16 | tmp$new_x <- as.character(new_x[i]) 17 | tmp$vname <- vname 18 | tmp$x_quant <- quant_x 19 | tmp$quant <- probs[i] 20 | tmp$relative_quant <- probs[i] - quant_x 21 | tmp$label <- explainer_s$label 22 | tmp$time <- times_s 23 | tmp$class <- "numeric" 24 | res <- rbind(res, tmp) 25 | } 26 | } 27 | if(class(data_s[,vname])=="character" || class(data_s[,vname])=="factor"){ 28 | data_s[,vname] <- as.factor(data_s[,vname]) 29 | new_data <- observation_s[rep(1, length(levels(data_s[,vname]))),] 30 | new_data[,vname] <- as.factor(new_data[,vname]) 31 | new_x <- levels(data_s[,vname]) 32 | new_data[,vname] <- new_x 33 | f <- sapply(data_s, is.factor) 34 | cols <- names(which(f)) 35 | new_data[cols] <- lapply(new_data[cols], as.factor) 36 | y_hat <- t(predict_function_s(model_s, new_data, times_s)) 37 | 38 | res <- data.frame(y_hat=numeric(), time = numeric(), vname = character(), new_x = character(), 39 | x_quant = numeric(), quant = numeric(), relative_quant = numeric(), label = character(), 40 | class = character()) 41 | 42 | for(i in 1:length(levels(data_s[,vname]))){ 43 | tmp <- data.frame(y_hat = y_hat[,i]) 44 | tmp$new_x <- new_x[i] 45 | tmp$vname <- vname 46 | tmp$x_quant <- 0 47 | tmp$quant <- 0 48 | tmp$relative_quant <- 0 49 | tmp$label <- explainer_s$label 50 | tmp$time <- times_s 51 | tmp$class <- "factor" 52 | res <- rbind(res, tmp) 53 | } 54 | } 55 | return(res) 56 | } 57 | -------------------------------------------------------------------------------- /R/ceteris_paribus.R: -------------------------------------------------------------------------------- 1 | #' Ceteris Paribus 2 | #' 3 | #' @description The \code{ceteris_paribus()} function computes the predictions for the neighbor of our chosen observation. The neighbour is defined as the observations with changed value of one of the variable. 4 | #' 5 | #' @param explainer a model to be explained, preprocessed by the 'survxai::explain' function 6 | #' @param observation a new observation for which predictions need to be explained 7 | #' @param grid_points grid_points number of points used for response path 8 | #' @param selected_variables if specified, then only these variables will be explained 9 | #' 10 | #' @return An object of the class surv_ceteris_paribus_explainer. 11 | #' It's a data frame with calculated average responses. 12 | #' @export 13 | #' 14 | #' @importFrom stats quantile 15 | #' @importFrom utils head 16 | #' 17 | #' @examples 18 | #' \donttest{ 19 | #' library(survxai) 20 | #' library(rms) 21 | #' data("pbcTrain") 22 | #' data("pbcTest") 23 | #' predict_times <- function(model, data, times){ 24 | #' prob <- rms::survest(model, data, times = times)$surv 25 | #' return(prob) 26 | #' } 27 | #' cph_model <- cph(Surv(years, status)~ sex + bili + stage, 28 | #' data = pbcTrain, surv = TRUE, x = TRUE, y=TRUE) 29 | #' surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 30 | #' y = Surv(pbcTest$years, pbcTest$status), 31 | #' predict_function = predict_times) 32 | #' cp_cph <- ceteris_paribus(surve_cph, pbcTest[1,-c(1,5)]) 33 | #' } 34 | #' @export 35 | 36 | ceteris_paribus <- function(explainer, observation, grid_points = 5, selected_variables = NULL){ 37 | if (!("surv_explainer" %in% class(explainer))) 38 | stop("The ceteris_paribus() function requires an object created with explain() function from survxai package.") 39 | if (is.null(explainer$data)) 40 | stop("The ceteris_paribus() function requires explainers created with specified 'data' parameter.") 41 | 42 | data <- base::as.data.frame(explainer$data) 43 | model <- explainer$model 44 | predict_function <- explainer$predict_function 45 | names_to_present <- colnames(data) 46 | grid_points <- grid_points 47 | 48 | if (!is.null(selected_variables)) { 49 | names_to_present <- intersect(names_to_present, selected_variables) 50 | } 51 | 52 | times <- explainer$times 53 | times <- sort(times) 54 | 55 | responses <- lapply(names_to_present, function(vname, times_s, observation_s, model_s, explainer_s, grid_points_s, data_s, predict_function_s) calculate_responses(vname,times_s = times, observation_s = observation, model_s = model, explainer_s = explainer, grid_points_s = grid_points, data_s = data, predict_function_s = predict_function)) 56 | 57 | all_responses <- do.call(rbind, responses) 58 | new_y_hat <- predict_function(model, observation, times) 59 | attr(all_responses, "prediction") <- list(observation = observation, new_y_hat = new_y_hat, times = times) 60 | attr(all_responses, "grid_points") <- grid_points 61 | 62 | class(all_responses) <- c("surv_ceteris_paribus_explainer", "data.frame") 63 | all_responses 64 | } 65 | -------------------------------------------------------------------------------- /R/explain.R: -------------------------------------------------------------------------------- 1 | #' @title Create Survival Model Explainer 2 | #' 3 | #' @description Survival models may have very different structures. 4 | #' This function creates a unified representation of a survival model, which can be further processed by various survival 5 | #' explainers (see also \code{\link[DALEX]{explain}}). 6 | #' 7 | #' Please NOTE, that the \code{model} is actually the only required argument. 8 | #' But some survival explainers may require additional arguments. 9 | #' 10 | #' @param model object - a survival model to be explained 11 | #' @param data data.frame, tibble or matrix - data that will be used by survival explainers. If not provided then will be extracted from the model 12 | #' @param y object of class 'surv', contains event status and times 13 | #' @param times optional argument, the vector of time points on which survival probability will be predicted 14 | #' @param predict_function function that takes three arguments: model, new data, vector with times, and returns numeric vector or matrix with predictions. If not passed, function \code{\link[pec]{predictSurvProb}} is used. 15 | #' @param link function - a transformation/link function that shall be applied to raw model predictions 16 | #' @param label character - the name of the survival model. By default it's extracted from the 'class' attribute of the model. 17 | #' @param ... other parameters 18 | #' 19 | #' @return An object of the class 'surv_explainer'. 20 | #' 21 | #' It's a list with following fields: 22 | #' 23 | #' \itemize{ 24 | #' \item \code{model} the explained model 25 | #' \item \code{data} the dataset 26 | #' \item \code{y} event statuses and times 27 | #' \item \code{times} time points on which survival probability is predicted 28 | #' \item \code{predict_function} function that may be used for model predictions, shall return a single numerical value for each time. 29 | #' \item \code{link} function - a transformation/link function that shall be applied to raw model predictions 30 | #' \item \code{class} class/classes of a model 31 | #' \item \code{label} label, by default it's the last value from the \code{class} vector, but may be set to any character. 32 | #' } 33 | #' 34 | #' @importFrom stats predict model.frame 35 | #' @importFrom utils head tail 36 | #' 37 | #' @examples 38 | #' \donttest{ 39 | #' library(survxai) 40 | #' library(rms) 41 | #' library(randomForestSRC) 42 | #' data(pbc, package = "randomForestSRC") 43 | #' pbc <- pbc[complete.cases(pbc),] 44 | #' predict_times <- function(model, data, times){ 45 | #' prob <- rms::survest(model, data, times = times)$surv 46 | #' return(prob) 47 | #' } 48 | #' cph_model <- cph(Surv(days/365, status)~ sex + bili + stage, data=pbc, surv=TRUE, x = TRUE, y=TRUE) 49 | #' surve_cph <- explain(model = cph_model, data = pbc[,-c(1,2)], y = Surv(pbc$days/365, pbc$status), 50 | #' predict_function = predict_times) 51 | #' } 52 | #' @export 53 | 54 | explain <- function(model, data = NULL, y, times = NULL, predict_function = yhat, link = I, label = tail(class(model), 1), ...) UseMethod("explain") 55 | 56 | #' @rdname explain 57 | #' @export 58 | explain.default <- function(model, data = NULL, y, times = NULL, predict_function = yhat, link = I, label = tail(class(model), 1), ...) { 59 | if (is.null(times)) times <- y[,1] 60 | 61 | if (is.null(data)) { 62 | possible_data <- try(model.frame(model), silent = TRUE) 63 | if (class(possible_data) != "try-error") data <- possible_data 64 | } 65 | # if data is in the tibble format then needs to be translated to data.frame 66 | if ("tbl" %in% class(data)) data <- as.data.frame(data) 67 | 68 | surv_explainer <- list( 69 | model = model, 70 | data = data, 71 | y = y, 72 | times = times, 73 | predict_function = predict_function, 74 | link = link, 75 | class = class(model), 76 | label = label 77 | ) 78 | surv_explainer <- c(surv_explainer, list(...)) 79 | class(surv_explainer) <- "surv_explainer" 80 | attr(surv_explainer, "formula") <- deparse(substitute(y)) 81 | return(surv_explainer) 82 | } 83 | 84 | 85 | 86 | #' @method predictSurvProb surv_explainer 87 | #' @export 88 | predictSurvProb.surv_explainer <- function(object, newdata, times, ...) { 89 | object$predict_function(object$model, newdata, times) 90 | } 91 | 92 | yhat <- function(X.model, newdata, times) { 93 | predictSurvProb(X.model, newdata, times) 94 | } 95 | -------------------------------------------------------------------------------- /R/model_performance.R: -------------------------------------------------------------------------------- 1 | #' @title Model performance for survival models 2 | #' 3 | #' @description Function \code{model_performance} calculates the prediction error for chosen survival model. 4 | #' 5 | #' @param explainer a model to be explained, preprocessed by the 'survxai::explain' function 6 | #' @param type character - type of the response to be calculated 7 | #' Currently following options are implemented: 'BS' for Expected Brier Score 8 | #' 9 | #' @details 10 | #' For \code{type = "BS"} prediction error is the time dependent estimates of the population average Brier score. 11 | #' At a given time point t, the Brier score for a single observation is the squared difference between observed survival status 12 | #' and a model based prediction of surviving time t. 13 | #' 14 | #' @examples 15 | #' \donttest{ 16 | #' library(survxai) 17 | #' library(rms) 18 | #' data("pbcTrain") 19 | #' data("pbcTest") 20 | #' cph_model <- cph(Surv(years, status)~ sex + bili + stage, 21 | #' data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 22 | #' surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 23 | #' y = Surv(pbcTest$years, pbcTest$status)) 24 | #' mp_cph <- model_performance(surve_cph) 25 | #' } 26 | #' 27 | #' @references Ulla B. Mogensen, Hemant Ishwaran, Thomas A. Gerds (2012). Evaluating Random Forests for Survival Analysis Using Prediction Error Curves. Journal of Statistical Software, 50(11), 1-23. URL http://www.jstatsoft.org/v50/i11/. 28 | #' 29 | #' @import pec 30 | #' @importFrom prodlim Hist 31 | #' @importFrom stats as.formula 32 | #' 33 | #' @export 34 | 35 | model_performance <- function(explainer, type = "BS"){ 36 | if (!("surv_explainer" %in% class(explainer))) stop("The model_performance() function requires an object created with explain() function from survxai package.") 37 | reference_formula <- eval(explainer$model$call[[2]]) 38 | # trick for mlr, to remove third param in Surv 39 | if(length(reference_formula[[2]]) > 3){ 40 | reference_formula[[2]][4] <- NULL 41 | } 42 | reference_formula[3] <- 1 43 | surv_vars <- all.vars(explainer$model$call[[2]][[2]]) 44 | data <- cbind(explainer$y[,1], explainer$y[,2], explainer$data) 45 | colnames(data)[1:2] <- surv_vars 46 | 47 | switch(type, 48 | BS = { 49 | p <- tryCatch({ 50 | p <- pec(explainer$model, data = data, splitMethod = "none", formula = reference_formula) 51 | }, error = function(e) { 52 | p <- pec(explainer, data = data, splitMethod = "none", formula = reference_formula, reference = TRUE) 53 | return(p) 54 | }) 55 | res <- data.frame(time = p$time, err = p$AppErr[[2]], err_ref = p$AppErr[[1]], label = explainer$label) 56 | class(res) <- c("surv_model_performance_explainer", "data.frame", "BS") 57 | attr(res, "type") <- type 58 | attr(res, "time") <- explainer$y[,1] 59 | return(res) 60 | }, 61 | stop("Currently only 'BS' method is implemented")) 62 | } 63 | 64 | 65 | 66 | 67 | -------------------------------------------------------------------------------- /R/pbcTest.R: -------------------------------------------------------------------------------- 1 | #' @title pbcTest 2 | #' @description PBC test set 3 | #' Data set based on \code{pbc} from \code{randomForestSRC} package. 4 | #' The data consists of 138 randomly chosen observations The \code{pbcTest} contains only complete cases for each observation. 5 | #' It contains 5 variables: `status`, `sex`, `bili`, `stage`, and `years`. 6 | #' 7 | #' @source randomForestSRC 8 | #' @references Flemming T.R and Harrington D.P., (1991) Counting Processes and Survival Analysis. New York: Wiley. 9 | #' @name pbcTest 10 | #' @docType data 11 | #' 12 | #' @examples 13 | #' data("pbcTest", package = "survxai") 14 | #' head(pbcTest) 15 | NULL 16 | -------------------------------------------------------------------------------- /R/pbcTrain.R: -------------------------------------------------------------------------------- 1 | #' @title pbcTrain 2 | #' @description PBC train set 3 | #' Data set based on \code{pbc} from \code{randomForestSRC} package. 4 | #' The data consists of 138 randomly chosen observations The \code{pbcTrain} contains only complete cases for each observation. 5 | #' It contains 5 variables: `status`, `sex`, `bili`, `stage`, and `years`. 6 | #' 7 | #' @source randomForestSRC 8 | #' @references Flemming T.R and Harrington D.P., (1991) Counting Processes and Survival Analysis. New York: Wiley. 9 | #' @name pbcTrain 10 | #' @docType data 11 | #' 12 | #' @examples 13 | #' data("pbcTrain", package = "survxai") 14 | #' head(pbcTrain) 15 | NULL 16 | -------------------------------------------------------------------------------- /R/plot_ceteris_paribus.R: -------------------------------------------------------------------------------- 1 | #' @title Plot for ceteris_paribus object 2 | #' 3 | #' @description Function plot for ceteris_paribus object visualise estimated survival curve of mean probabilities in chosen time points. Black lines on each plot correspond to survival curve for our new observation specified in the \code{ceteris_paribus} function. 4 | #' 5 | #' @param x object of class "surv_ceteris_paribus_explainer" 6 | #' @param ... arguments to be passed to methods, such as graphical parameters for function \code{\link[ggplot2]{geom_step}}. 7 | #' @param selected_variable name of variable we want to draw ceteris paribus plot 8 | #' @param scale_type type of scale of colors, either "discrete" or "gradient" 9 | #' @param scale_col vector containing values of low and high ends of the gradient, when "gradient" type of scale was chosen 10 | #' @param ncol number of columns for faceting 11 | #' 12 | #' @import ggplot2 13 | #' @importFrom scales seq_gradient_pal 14 | #' @examples 15 | #' \donttest{ 16 | #' library(survxai) 17 | #' library(rms) 18 | #' data("pbcTest") 19 | #' data("pbcTrain") 20 | #' predict_times <- function(model, data, times){ 21 | #' prob <- rms::survest(model, data, times = times)$surv 22 | #' return(prob) 23 | #' } 24 | #' cph_model <- cph(Surv(years, status)~sex + bili + stage, data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 25 | #' surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 26 | #' y = Surv(pbcTest$years, pbcTest$status), predict_function = predict_times) 27 | #' cp_cph <- ceteris_paribus(surve_cph, pbcTest[1,-c(1,5)]) 28 | #' plot(cp_cph) 29 | #' } 30 | #' @method plot surv_ceteris_paribus_explainer 31 | #' @export 32 | 33 | plot.surv_ceteris_paribus_explainer <- function(x, ..., selected_variable = NULL, scale_type = "factor", 34 | scale_col = NULL, ncol = 1) { 35 | 36 | if(!is.null(selected_variable) && !(selected_variable %in% factor(x$vname))){ 37 | stop(paste0("Selected variable ", selected_variable, "not present in surv_ceteris_paribus object.")) 38 | } 39 | 40 | y_hat <- new_x <- time <- time_2 <- y_hat_2 <- NULL 41 | new_observation_legend <- create_legend(x=x) 42 | seq_length <- attributes(x)$grid_points 43 | 44 | all_responses <- x 45 | 46 | all_predictions <- create_predictions(x) 47 | 48 | 49 | all_responses <- merge(all_responses, new_observation_legend, by="vname") 50 | if(!is.null(selected_variable)){ 51 | all_responses <- all_responses[which(all_responses$vname == selected_variable),] 52 | legend <- unique(all_responses$val) 53 | add_theme <- labs(col = legend) 54 | facet <- NULL 55 | title <- ggtitle(paste("Ceteris paribus plot for variable", selected_variable,".")) 56 | }else{ 57 | add_theme <- theme(legend.position = "none") 58 | title <- ggtitle(paste("Ceteris paribus plot for", unique(x$label),"model.")) 59 | facet <- facet_wrap(~val, ncol = ncol) 60 | } 61 | 62 | ####################### 63 | df <- all_responses[,c("vname","new_x")] 64 | df <- unique(df) 65 | df$legend <- 1:nrow(df) 66 | all_responses <- merge(all_responses, df, by=c("vname", "new_x")) 67 | 68 | ############################ 69 | scale <- create_scale(all_responses, scale_type, scale_col, selected_variable) 70 | 71 | ggplot(all_responses, aes(x = time, y = y_hat, col = factor(legend))) + 72 | geom_step(...) + 73 | geom_step(data = all_predictions, aes(x = time_2, y = y_hat_2,...), col="black", lty = 2, size = 1) + 74 | scale_y_continuous(breaks = seq(0,1,0.1), 75 | limits = c(0,1), 76 | labels = paste(seq(0,100,10),"%"), 77 | name = "survival probability") + 78 | facet + 79 | theme_mi2() + 80 | add_theme + 81 | title + 82 | scale 83 | } 84 | 85 | create_legend <- function(x){ 86 | new_observation <- attributes(x)$prediction$`observation` 87 | values <- as.data.frame(t(new_observation[1,])) 88 | values[,1] <- as.character(values[,1]) 89 | new_observation_legend <- data.frame(vname = colnames(new_observation), val = paste0(colnames(new_observation), "=", values[,1])) 90 | return(new_observation_legend) 91 | } 92 | 93 | 94 | create_predictions <- function(x){ 95 | pred <- attr(x, "prediction") 96 | all_predictions <- data.frame(prediction = pred$new_y_hat) 97 | times <- data.frame(prediction = pred$times) 98 | all_predictions <- data.frame(t(all_predictions)) 99 | all_predictions$time_2 <- times$prediction 100 | colnames(all_predictions)[1] <- "y_hat_2" 101 | return(all_predictions) 102 | } 103 | 104 | 105 | create_scale <- function(all_responses, scale_type, scale_col, selected_variable){ 106 | if(scale_type == "gradient"){ 107 | if(!is.null(scale_col)){ 108 | variables <- unique(all_responses$vname) 109 | v<- c() 110 | for(val in variables){ 111 | length <- length(unique(all_responses[all_responses$vname==val,2])) 112 | cc <- seq_gradient_pal(scale_col[1],scale_col[2])(seq(0,1,length.out=length)) 113 | v <- c(v,cc) 114 | } 115 | if(!is.null(selected_variable)){ 116 | scale <- scale_colour_manual(values = v, labels = factor(unique(all_responses$new_x))) 117 | }else{ 118 | scale <- scale_colour_manual(values=v) 119 | } 120 | }else{ 121 | message("Please specify the low and high ends of gradient") 122 | scale <- NULL 123 | } 124 | }else{ 125 | scale <- NULL 126 | } 127 | return(scale) 128 | } 129 | -------------------------------------------------------------------------------- /R/plot_explainer.R: -------------------------------------------------------------------------------- 1 | #' @title Plot for surv_explainer object 2 | #' 3 | #' @description Function plot for surv_explainer object visualise estimated survival curve of mean probabilities in chosen time points. 4 | #' 5 | #' @param x object of class "surv_explainer" 6 | #' @param ... other arguments for function \code{\link[survminer]{ggsurvplot}} 7 | #' 8 | #' @import ggplot2 9 | #' @importFrom survival survfit 10 | #' @importFrom survminer ggsurvplot 11 | #' @examples 12 | #' \donttest{ 13 | #' library(survxai) 14 | #' library(rms) 15 | #' data("pbcTest") 16 | #' data("pbcTrain") 17 | #' predict_times <- function(model, data, times){ 18 | #' prob <- rms::survest(model, data, times = times)$surv 19 | #' return(prob) 20 | #' } 21 | #' cph_model <- cph(Surv(years, status)~sex + bili + stage, data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 22 | #' surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 23 | #' y = Surv(pbcTest$years, pbcTest$status), predict_function = predict_times) 24 | #' plot(surve_cph) 25 | #' } 26 | #' @method plot surv_explainer 27 | #' @export 28 | 29 | 30 | plot.surv_explainer <- function(x, ...){ 31 | fit <- survfit(x$model, data = x$data) 32 | ggsurvplot(fit, data = x$data,...) 33 | 34 | } 35 | -------------------------------------------------------------------------------- /R/plot_model_performance.R: -------------------------------------------------------------------------------- 1 | #' @title Plot for surv_model_performance object 2 | #' 3 | #' @description Function plot for surv_model_performance object. 4 | #' 5 | #' @param x object of class "surv_model_performance" 6 | #' @param ... optional, additional objects of class "surv_model_performance_explainer" 7 | #' 8 | #' @import ggplot2 9 | #' 10 | #' @examples 11 | #' \donttest{ 12 | #' library(survxai) 13 | #' library(rms) 14 | #' data("pbcTest") 15 | #' data("pbcTrain") 16 | #' predict_times <- function(model, data, times){ 17 | #' prob <- rms::survest(model, data, times = times)$surv 18 | #' return(prob) 19 | #' } 20 | #' cph_model <- cph(Surv(years, status)~sex + bili + stage, data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 21 | #'surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 22 | #' y = Surv(pbcTest$years, pbcTest$status), predict_function = predict_times) 23 | #' mp_cph <- model_performance(surve_cph) 24 | #' plot(mp_cph) 25 | #' } 26 | #' 27 | #' @method plot surv_model_performance_explainer 28 | #' @export 29 | 30 | plot.surv_model_performance_explainer <- function(x, ...){ 31 | time <- err <- label <- NULL 32 | 33 | df <- data.frame(x) 34 | type <- attributes(x)$type 35 | if (type == "BS") type <- "Brier Score" 36 | 37 | dfl <- list(...) 38 | if (length(dfl) > 0) { 39 | for (resp in dfl) { 40 | class(resp) <- "data.frame" 41 | df <- rbind(df, resp) 42 | } 43 | } 44 | 45 | 46 | ggplot(df, aes(x = time, y = err, color = label)) + 47 | geom_step() + 48 | labs(title = paste("Prediction Error Curve for", type,"method"), 49 | x = "time", 50 | y = "prediction error") + 51 | theme_mi2()+ 52 | scale_y_continuous(breaks = seq(0,1,0.1), 53 | limits = c(0,1), 54 | labels = paste(seq(0,100,10),"%")) 55 | 56 | 57 | } 58 | -------------------------------------------------------------------------------- /R/plot_prediction_breakdown.R: -------------------------------------------------------------------------------- 1 | #' @title Plot for surv_breakdown object 2 | #' 3 | #' @description Function plot for surv_breakdown object visualise estimated survival curve of mean probabilities in chosen time points. 4 | #' 5 | #' @param x an object of class "surv_prediction_breakdown_explainer" 6 | #' @param ... optional, additional objects of class "surv_prediction_breakdown_explainer" 7 | #' @param numerate logical; indicating whether we want to number curves 8 | #' @param lines logical; indicating whether we want to add lines on chosen time point or probability 9 | #' @param lines_type a type of line; see http://sape.inf.usi.ch/quick-reference/ggplot2/linetype 10 | #' @param lines_col a color of line 11 | #' @param scale_col a vector containig two colors for gradient scale in legend 12 | #' 13 | #' 14 | #' @import ggplot2 15 | #' @examples 16 | #' \donttest{ 17 | #' library(survxai) 18 | #' library(rms) 19 | #' data("pbcTest") 20 | #' data("pbcTrain") 21 | #' predict_times <- function(model, data, times){ 22 | #' prob <- rms::survest(model, data, times = times)$surv 23 | #' return(prob) 24 | #' } 25 | #' cph_model <- cph(Surv(years, status)~sex + bili + stage, data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 26 | #' surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 27 | #' y = Surv(pbcTest$years, pbcTest$status), predict_function = predict_times) 28 | #' broken_prediction <- prediction_breakdown(surve_cph, pbcTest[1,-c(1,5)]) 29 | #' plot(broken_prediction) 30 | #' } 31 | #' @method plot surv_prediction_breakdown_explainer 32 | #' 33 | #' @importFrom scales seq_gradient_pal 34 | #' @export 35 | 36 | plot.surv_prediction_breakdown_explainer <- function(x, ..., numerate = TRUE, lines = TRUE, 37 | lines_type = 1, lines_col = "black", 38 | scale_col = c("#010059","#e0f6fb")){ 39 | y <- col <- label <- value <- position <- legend <- NULL 40 | 41 | df <- data.frame(x) 42 | dfl <- list(...) 43 | if (length(dfl) > 0) { 44 | for (resp in dfl) { 45 | class(resp) <- "data.frame" 46 | df <- rbind(df, resp) 47 | } 48 | } 49 | 50 | if(length(dfl)==0){ 51 | add_facet <- NULL 52 | legend <- NULL 53 | }else{ 54 | add_facet <- facet_wrap(~label, ncol = 1) 55 | legend <- theme(legend.position = "none") 56 | } 57 | 58 | 59 | df <- create_legend_broken(df, x) 60 | #colors 61 | cc <- seq_gradient_pal(scale_col[1],scale_col[2])(seq(0,1,length.out=length(unique(df$legend)))) 62 | 63 | median_time <- median(unique(df$x)) 64 | median <- which.min(abs(unique(df$x) - median_time)) 65 | median <- unique(df$x)[median] 66 | 67 | if(!is.null(attributes(x)$prob)){ 68 | line <- geom_hline(yintercept = attributes(x)$prob, color = lines_col, linetype = lines_type) 69 | }else if (!is.null(attributes(x)$time)){ 70 | line <- geom_vline(xintercept = attributes(x)$time, color = lines_col, linetype = lines_type) 71 | }else{ 72 | line <- geom_vline(xintercept = median, color = lines_col, linetype = lines_type) 73 | } 74 | 75 | if(lines == TRUE){ 76 | line <- line 77 | }else{ 78 | line <- NULL 79 | } 80 | 81 | if(numerate == TRUE){ 82 | numbers <- geom_text(data = df[df$x == median,], aes(label = position), color = "black", show.legend = FALSE, hjust = 0, vjust = 0, nudge_x = 0.4) 83 | }else{ 84 | numbers <- NULL 85 | } 86 | 87 | ggplot(df, aes(x=x, y=y, col = factor(legend)))+ 88 | geom_step()+ 89 | numbers+ 90 | labs(title = "BreakDown plot", 91 | x = "time", 92 | y = "mean survival probability", 93 | col = "variable") + 94 | add_facet + 95 | theme_mi2()+ 96 | scale_colour_manual(values=cc)+ 97 | line+ 98 | numbers+ 99 | scale_y_continuous(breaks = seq(0,1,0.1), 100 | limits = c(0,1), 101 | labels = paste(seq(0,100,10),"%"), 102 | name = "survival probability") + 103 | legend 104 | 105 | 106 | } 107 | 108 | create_legend_broken <- function(df, x){ 109 | df$legend <- paste0(df$position,": ", df$value) 110 | broken_cumm <- attributes(x)$contribution 111 | broken_cumm$contribution <- round(broken_cumm$contribution, digits = 2) 112 | broken_cumm$contribution <- paste0("(", broken_cumm$contribution, ")") 113 | broken_cumm$variable <- as.character(broken_cumm$variable) 114 | broken_cumm <- rbind(broken_cumm, c(round(attributes(x)$Intercept,2), "Intercept")) 115 | broken_cumm <- rbind(broken_cumm, c(round(attributes(x)$Observation,2), "Observation")) 116 | 117 | df <- merge(df, broken_cumm, by = "variable") 118 | df$legend <- paste(df$legend, df$contribution) 119 | df$legend <- factor(df$legend, levels = unique(df$legend[order(df$position)])) 120 | return(df) 121 | } 122 | -------------------------------------------------------------------------------- /R/plot_variable_response.R: -------------------------------------------------------------------------------- 1 | #' @title Plot for surv_variable_response object 2 | #' 3 | #' @description Function plot for surv_variable_response object shows the expected output condition on a selected variable. 4 | #' 5 | #' @param x an object of class "surv_variable_response" 6 | #' @param ... optional, additional objects of class "surv_variable_response_explainer" 7 | #' @param split a character, either "model" or "variable"; sets the variable for faceting 8 | #' 9 | #' @import ggplot2 10 | #' @importFrom stats aggregate quantile 11 | #' 12 | #' @examples 13 | #' \donttest{ 14 | #' library(survxai) 15 | #' library(rms) 16 | #' data("pbcTest") 17 | #' data("pbcTrain") 18 | #' predict_times <- function(model, data, times){ 19 | #' prob <- rms::survest(model, data, times = times)$surv 20 | #' return(prob) 21 | #' } 22 | #' cph_model <- cph(Surv(years, status)~sex + bili + stage, data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 23 | #' surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 24 | #' y = Surv(pbcTest$years, pbcTest$status), predict_function = predict_times) 25 | #' svr_cph <- variable_response(surve_cph, "sex") 26 | #' plot(svr_cph) 27 | #' } 28 | #' 29 | #' @method plot surv_variable_response_explainer 30 | #' @export 31 | 32 | plot.surv_variable_response_explainer <- function(x, ..., split = "model"){ 33 | y <- color <- NULL 34 | 35 | df <- data.frame(x) 36 | dfl <- list(...) 37 | if (length(dfl) > 0) { 38 | for (resp in dfl) { 39 | class(resp) <- "data.frame" 40 | df <- rbind(df, resp) 41 | } 42 | } 43 | 44 | if (is.numeric(df$value) & length(unique(df$value))>=4) { 45 | df$value <- cut(df$value, quantile(df$value, prob = seq(0, 1, length.out = 6)), include.lowest = TRUE) 46 | df <- aggregate(y~., data = df, mean) 47 | } 48 | 49 | if (split == "variable") { 50 | add_facet <- facet_wrap(~value, ncol = 1) 51 | df$color <- factor(df$label) 52 | legend <- "model" 53 | } else { 54 | add_facet <- facet_wrap(~label, ncol = 1) 55 | df$color <- factor(df$value) 56 | legend <- x$var[1] 57 | } 58 | 59 | 60 | ggplot(df, aes(x, y, color = color)) + 61 | geom_step() + 62 | labs(title = paste0("Partial Dependency Plot of variable ", df$var[1]), 63 | x = "time", 64 | y = "mean survival probability", 65 | col = legend) + 66 | add_facet + 67 | theme_mi2()+ 68 | scale_y_continuous(breaks = seq(0,1,0.1), 69 | limits = c(0,1), 70 | labels = paste(seq(0,100,10),"%"), 71 | name = "survival probability") 72 | 73 | 74 | } 75 | -------------------------------------------------------------------------------- /R/prediction_breakdown.R: -------------------------------------------------------------------------------- 1 | #' @title BreakDown for survival models 2 | #' 3 | #' @description Function \code{prediction_breakdown} is an extension of a broken function from breakDown package. It computes the contribution in prediction for the variables in the model. 4 | #' The contribution is defined as the difference between survival probabilities for model with added specific value of variable and with the random levels of this variable. 5 | #' 6 | #' @param explainer an object of the class 'surv_explainer' 7 | #' @param observation a new observation to explain 8 | #' @param time a time point at which variable contributions are computed. If NULL median time is taken. 9 | #' @param prob a survival probability at which variable contributions are computed 10 | #' @param ... other parameters corresponding to arguments from \code{\link[breakDown]{broken}} function from \code{breakDown} package. See https://github.com/pbiecek/breakDown/blob/master/R/break_agnostic.R for details 11 | #' 12 | #' @return An object of class surv_prediction_breakdown_explainer 13 | #' 14 | #' @importFrom breakDown broken 15 | #' @importFrom stats weighted.mean na.omit median 16 | #' 17 | #' @examples 18 | #' \donttest{ 19 | #' library(survxai) 20 | #' library(rms) 21 | #' data("pbcTest") 22 | #' data("pbcTrain") 23 | #' predict_times <- function(model, data, times){ 24 | #' prob <- rms::survest(model, data, times = times)$surv 25 | #' return(prob) 26 | #' } 27 | #' cph_model <- cph(Surv(years, status)~sex + bili + stage, data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 28 | #' surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 29 | #' y = Surv(pbcTest$years, pbcTest$status), predict_function = predict_times) 30 | #' broken_prediction <- prediction_breakdown(surve_cph, pbcTest[1,-c(1,5)]) 31 | #' } 32 | #' @export 33 | 34 | 35 | prediction_breakdown <- function(explainer, observation, time = NULL, prob = NULL, ...){ 36 | if (!("surv_explainer" %in% class(explainer))) stop("The prediction_breakdown() function requires an object created with explain() function form survxai package.") 37 | if (is.null(explainer$data)) stop("The prediction_breakdown() function requires explainers created with specified 'data' parameter.") 38 | if (!is.null(time) & !is.null(prob)) stop("Only one of the parameters 'time', 'prob' should be provided.") 39 | 40 | # breakDown 41 | new_pred <- predict_fun(prob, time, explainer) 42 | 43 | oldw <- getOption("warn") 44 | options(warn = -1) 45 | res<- broken(model = explainer$model, 46 | new_observation = observation, 47 | data = explainer$data, 48 | predict.function = new_pred, 49 | ...) 50 | options(warn = oldw) 51 | 52 | class(res) <- "data.frame" 53 | 54 | intercept <- res$contribution[res$variable_name=="Intercept"] 55 | observ <- res$contribution[res$variable=="final_prognosis"] 56 | 57 | result <- data.frame(x = numeric(), y = numeric(), variable = character(), label = character(), position = numeric(), value = character()) 58 | res <- res[-c(1, nrow(res)),] 59 | 60 | 61 | times <- sort(explainer$times) 62 | 63 | #baseline 64 | mean_prediction <- calculate_prediction_intercept(explainer, times) 65 | result <- rbind(result, mean_prediction) 66 | #one observation 67 | mean_prediction <- calculate_prediction_observation(explainer, observation, times, res) 68 | result <- rbind(result, mean_prediction) 69 | tmp_data <- explainer$data 70 | 71 | for (i in 1:nrow(res)){ 72 | #explainer <- explainer 73 | variable <- res[i, "variable_name"] 74 | tmp_data[,as.character(variable)] <- observation[[as.character(variable)]] 75 | mean_prediction <- calculate_prediction(explainer, tmp_data, times, res, i, variable) 76 | result <- rbind(result, mean_prediction) 77 | } 78 | 79 | res <- res[,c("contribution", "variable_name")] 80 | colnames(res)[2] <- "variable" 81 | attr(result, "contribution") <- res 82 | attr(result, "time") <- time 83 | attr(result, "prob") <- prob 84 | attr(result, "Intercept") <- intercept 85 | attr(result, "Observation") <- observ 86 | class(result) <- c("surv_prediction_breakdown_explainer", "data.frame") 87 | result 88 | 89 | } 90 | 91 | 92 | predict_fun <- function(prob, time, explainer){ 93 | if (is.null(prob)) { 94 | if (is.null(time)) time <- median(explainer$times) 95 | 96 | new_pred <- function(model, data){ 97 | explainer$predict_function(model, data, times = time) 98 | } 99 | } else { 100 | times_sorted <- sort(explainer$times) 101 | 102 | find_time <- function(x){ 103 | tim <- (x < prob) 104 | index <- c(min(which(tim == TRUE)) -1, min(which(tim == TRUE))) 105 | closest_times <- times_sorted[index] 106 | weighted.mean(closest_times, x[index]) 107 | } 108 | 109 | new_pred <- function(model, data){ 110 | probabilities <- explainer$predict_function(model, data, times = explainer$times) 111 | probabilities <- as.data.frame(probabilities) 112 | 113 | res <- apply(probabilities, MARGIN = 1, FUN = find_time) 114 | res <- na.omit(res) 115 | return(res) 116 | 117 | } 118 | 119 | npred <- new_pred(explainer$model, explainer$data) 120 | message("Number of observations with prob > ", prob, ": ", nrow(explainer$data) - length(npred)) 121 | } 122 | 123 | return(new_pred) 124 | } 125 | 126 | 127 | 128 | calculate_prediction_intercept <- function(explainer, times){ 129 | prediction <- explainer$predict_function(explainer$model, explainer$data, times) 130 | mean_prediction <- data.frame(x = times, y = colMeans(prediction, na.rm=T)) 131 | mean_prediction <- rbind(mean_prediction, c(0, 1)) 132 | mean_prediction$variable<- "Intercept" 133 | mean_prediction$label <- explainer$label 134 | mean_prediction$position <- 1 135 | mean_prediction$value <- "Intercept" 136 | return(mean_prediction) 137 | } 138 | 139 | calculate_prediction_observation <- function(explainer, observation, times, res){ 140 | prediction <- explainer$predict_function(explainer$model, observation, times) 141 | mean_prediction <- data.frame(x = times, y = prediction[1,]) 142 | mean_prediction <- rbind(mean_prediction, c(0, 1)) 143 | mean_prediction$variable<- "Observation" 144 | mean_prediction$label <- explainer$label 145 | mean_prediction$position <- nrow(res)+2 146 | mean_prediction$value <- "Observation" 147 | return(mean_prediction) 148 | } 149 | 150 | 151 | calculate_prediction <- function(explainer, tmp_data, times, res, i, variable){ 152 | prediction <- explainer$predict_function(explainer$model, tmp_data, times) 153 | mean_prediction <- data.frame(x = times, y = colMeans(prediction, na.rm=T)) 154 | mean_prediction <- rbind(mean_prediction, c(0, 1)) 155 | mean_prediction$variable<- variable 156 | mean_prediction$label <- explainer$label 157 | mean_prediction$position <- res[i, "position"] 158 | mean_prediction$value <- res[i, "variable"] 159 | return(mean_prediction) 160 | } 161 | -------------------------------------------------------------------------------- /R/print_ceteris_paribus.R: -------------------------------------------------------------------------------- 1 | #' Ceteris Paribus Print 2 | #' 3 | #' @param x the model of 'surv_ceteris_paribus_explainer' class 4 | #' @param ... further arguments passed to or from other methods 5 | #' 6 | #' @return a data frame 7 | #' 8 | #' @export 9 | 10 | print.surv_ceteris_paribus_explainer <- function(x, ...){ 11 | class(x) <- "data.frame" 12 | print(head(x, ...)) 13 | } -------------------------------------------------------------------------------- /R/print_explainer.R: -------------------------------------------------------------------------------- 1 | #' Print Survival Explainer Summary 2 | #' 3 | #' @param x a model survival expaliner created with the `explain()` function 4 | #' @param ... further arguments passed to or from other methods 5 | #' 6 | #' @export 7 | 8 | print.surv_explainer <- function(x, ...) { 9 | cat("Model label: ", x$label, "\n") 10 | cat("Model class: ", paste(x$class, collapse = ","), "\n") 11 | cat("Data head :\n") 12 | print(head(x$data,2)) 13 | return(invisible(NULL)) 14 | } -------------------------------------------------------------------------------- /R/print_model_performance.R: -------------------------------------------------------------------------------- 1 | #' Print Survival Model Performance 2 | #' 3 | #' @param x a model to be explained, object of the class 'model_performance_explainer' 4 | #' @param times a vector of integer times on which we want to check the value of prediction error 5 | #' @param ... further arguments passed to or from other methods 6 | #' 7 | #' @export 8 | 9 | print.surv_model_performance_explainer <- function(x, times = NULL, ...) { 10 | if (is.null(times)) times <- sort(attributes(x)$time)[1:10] 11 | 12 | x <- as.data.frame(x) 13 | x <- x[!duplicated(x$time),] 14 | x <- x[which(x$time %in% times),] 15 | rownames(x) <- NULL 16 | colnames(x)[2] <- "prediction error" 17 | x$`prediction error` <- x$`prediction error` * 100 18 | x$`prediction error` <- round(x$`prediction error`, digits = 2) 19 | x$`prediction error` <- paste0("~ ", x$`prediction error`, "%") 20 | 21 | type <- attributes(x)$type 22 | if (type == "BS") type <- "Brier Score" 23 | 24 | 25 | cat(paste("Model performance for", type, "method.")) 26 | cat("\n") 27 | print(x[,c("time","prediction error")]) 28 | } 29 | -------------------------------------------------------------------------------- /R/print_prediction_breakdown.R: -------------------------------------------------------------------------------- 1 | #' Prediction Breakdown Print 2 | #' 3 | #' @param x the model model of 'surv_prediction_breakdown_explainer' class 4 | #' @param ... further arguments passed to or from other methods 5 | #' @param digits number of decimal places (round) or significant digits (signif) to be used 6 | #' See the \code{rounding_function} argument 7 | #' @param rounding_function function that is to used for rounding numbers. 8 | #' It may be \code{signif()} which keeps a specified number of significant digits. 9 | #' Or the default \code{round()} to have the same precision for all components 10 | #' 11 | #' @export 12 | print.surv_prediction_breakdown_explainer <- function(x, ..., digits = 3, rounding_function = round) { 13 | broken_cumm <- attributes(x)$contribution 14 | class(broken_cumm) = "data.frame" 15 | broken_cumm$contribution <- broken_cumm$contribution*100 16 | broken_cumm$contribution <- rounding_function(broken_cumm$contribution, digits) 17 | broken_cumm <- broken_cumm[which(abs(broken_cumm$contribution)>=0.01),] 18 | broken_cumm$contribution <- paste0(broken_cumm$contribution, "%") 19 | print(broken_cumm[, "contribution", drop=FALSE]) 20 | } 21 | -------------------------------------------------------------------------------- /R/print_variable_response.R: -------------------------------------------------------------------------------- 1 | #' Variable Response Print 2 | #' 3 | #' @param x the model of 'surv_variable_response_explainer' class 4 | #' @param ... further arguments passed to or from other methods 5 | #' 6 | #' @return a data frame 7 | #' 8 | #' @export 9 | 10 | print.surv_variable_response_explainer <- function(x, ...){ 11 | class(x) <- "data.frame" 12 | print(head(x, ...)) 13 | } -------------------------------------------------------------------------------- /R/theme_mi2.R: -------------------------------------------------------------------------------- 1 | #' @title MI^2 plot theme 2 | #' 3 | #' @description ggplot theme for charts generated with MI^2 Data Lab packages. 4 | #' 5 | #' @return theme object that can be added to ggplot2 plots 6 | #' 7 | #' @export 8 | #' 9 | theme_mi2 <- function() { 10 | theme( 11 | axis.ticks = element_line(linetype = "blank"), 12 | axis.title = element_text(family = "sans"), 13 | plot.title = element_text(family = "sans"), 14 | legend.text = element_text(family = "sans"), 15 | legend.title = element_text(family = "sans"), 16 | panel.background = element_rect(fill = "#f5f5f5"), 17 | plot.background = element_rect( 18 | fill = "#f5f5f5", 19 | colour = "aliceblue", 20 | size = 0.8, 21 | linetype = "dotted" 22 | ), 23 | strip.background = element_rect(fill = "gray50"), 24 | strip.text = element_text(family = "sans"), 25 | legend.key = element_rect(fill = NA, colour = NA, size = 0), 26 | legend.background = element_rect(fill = NA) 27 | ) 28 | } 29 | -------------------------------------------------------------------------------- /R/variable_response.R: -------------------------------------------------------------------------------- 1 | #' @title Variable response for survival models 2 | #' 3 | #' @description Function \code{variable_response} calculates the expected output condition on a selected variable. 4 | #' 5 | #' @param explainer an object of the class 'surv_explainer'. 6 | #' @param variable a character with variable name. 7 | #' @param type a character - type of the response to be calculated. 8 | #' Currently following options are implemented: 'pdp' for Partial Dependency. 9 | #' @param link a function - a link function that shall be applied to raw model predictions. This will be inherited from the explainer. 10 | #' 11 | #' @examples 12 | #' \donttest{ 13 | #' library(survxai) 14 | #' library(rms) 15 | #' data("pbcTest") 16 | #' data("pbcTrain") 17 | #' predict_times <- function(model, data, times){ 18 | #' prob <- rms::survest(model, data, times = times)$surv 19 | #' return(prob) 20 | #' } 21 | #' cph_model <- cph(Surv(years, status)~sex + bili + stage, data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 22 | #' surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 23 | #' y = Surv(pbcTest$years, pbcTest$status), predict_function = predict_times) 24 | #' svr_cph <- variable_response(surve_cph, "sex") 25 | #' } 26 | #' @export 27 | 28 | variable_response <- function(explainer, variable, type = "pdp", link = explainer$link){ 29 | if (!("surv_explainer" %in% class(explainer))) stop("The variable_response() function requires an object created with explain() function from survxai package.") 30 | if (is.null(explainer$data)) stop("The variable_response() function requires explainers created with specified 'data' parameter.") 31 | 32 | switch(type, 33 | pdp = { 34 | res <- surv_partial(explainer, variable) 35 | class(res) <- c("surv_variable_response_explainer", "data.frame", "pdp") 36 | res 37 | }, 38 | stop("Currently only 'pdp' method is implemented")) 39 | } 40 | 41 | 42 | 43 | 44 | surv_partial <- function(explainer, variable){ 45 | times <- sort(explainer$times) 46 | tmp_data <- explainer$data 47 | values <- unique(explainer$data[,variable]) 48 | 49 | partial_data <- data.frame(x = numeric(), y = numeric(), value = character()) 50 | 51 | for(i in 1:length(values)){ 52 | val <- values[i] 53 | tmp_data[,variable] <- val 54 | prediction <- explainer$predict_function(explainer$model, tmp_data, times) 55 | mean_prediction <- data.frame(x = times, y = colMeans(prediction, na.rm=T)) 56 | mean_prediction <- rbind(mean_prediction, c(0, 1)) 57 | mean_prediction$value <- val 58 | partial_data <- rbind(partial_data, mean_prediction) 59 | } 60 | partial_data$type <- "pdp" 61 | partial_data$label <- explainer$label 62 | partial_data$var <- variable 63 | return(partial_data) 64 | } 65 | 66 | 67 | -------------------------------------------------------------------------------- /R/welcome.R: -------------------------------------------------------------------------------- 1 | .onAttach <- function(...) { 2 | packageStartupMessage(paste0("Welcome to survxai (version: ", utils::packageVersion("survxai"), ").", "\n","Information about the package can be found in the GitHub repository: https://github.com/MI2DataLab/survxai")) 3 | } -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | ```diff 2 | ! Please note that the `survxai` package is not actively maintained. 3 | + Consider using the succeeding `survex` package. 4 | ``` 5 | `survex` is available at https://github.com/modeloriented/survex. 6 | 7 | # survxai 8 | [![CRAN_Status_Badge](http://www.r-pkg.org/badges/version/survxai)](https://CRAN.R-project.org/package=survxai) 9 | [![Total Downloads](http://cranlogs.r-pkg.org/badges/grand-total/survxai)](http://cranlogs.r-pkg.org/badges/grand-total/survxai) 10 | [![Build Status](https://travis-ci.org/MI2DataLab/survxai.svg?branch=master)](https://travis-ci.org/MI2DataLab/survxai) 11 | [![Coverage Status](https://img.shields.io/codecov/c/github/MI2DataLab/survxai/master.svg)](https://codecov.io/github/MI2DataLab/survxai?branch=master) 12 | [![status](http://joss.theoj.org/papers/dcc9d53e8a1b1f613d59b9658b113fff/status.svg)](http://joss.theoj.org/papers/dcc9d53e8a1b1f613d59b9658b113fff) 13 | [![CII Best Practices](https://bestpractices.coreinfrastructure.org/projects/2123/badge)](https://bestpractices.coreinfrastructure.org/projects/2123) 14 | [![DOI](https://zenodo.org/badge/137778994.svg)](https://zenodo.org/badge/latestdoi/137778994) 15 | 16 | Survival analysis models are used primarily in medicine and churn analysis. Due to many applications, we are witnessing a fast development of a wide range of black-box survival models. Their lack of interpretability makes them unusable for analyzes that require an understanding of the model behavior. 17 | 18 | An R package survxai is a tool for creating explanations of survival models. For both, complex and simple survival models. It also enables to compare them. Currently, four explanation methods are implemented. We can divide them into 2 groups: local and global. 19 | 20 | The read more about the `surxvai` package see paper [survxai: an R package for structure-agnostic explanations of survival models](https://joss.theoj.org/papers/dcc9d53e8a1b1f613d59b9658b113fff) in The Journal of Open Source Software. 21 | 22 | ## Install 23 | ``` 24 | devtools::install_github("MI2DataLab/survxai") 25 | ``` 26 | 27 | ## CheatSheet 28 | 29 | 30 | 31 | 32 | 33 | ## News 34 | Informations about changes in `survxai` relases can be found in [NEWS](https://github.com/MI2DataLab/survxai/blob/master/NEWS.md). 35 | 36 | ## How to contribute 37 | Informations about creating issues and pull requests in `survxai` can be found in [CONTRIBUTING](https://github.com/MI2DataLab/survxai/blob/master/CONTRIBUTING.md). 38 | 39 | 40 | ## Acknowledgments 41 | Work on this package is financially supported by the 'NCN Opus grant 2016/21/B/ST6/02176'. 42 | -------------------------------------------------------------------------------- /_pkgdown.yml: -------------------------------------------------------------------------------- 1 | template: 2 | package: MI2template 3 | default_assets: false 4 | home: 5 | links: 6 | - text: Source code 7 | href: https://github.com/MI2DataLab/survxai 8 | - text: Contributing guidelines 9 | href: https://github.com/MI2DataLab/survxai/blob/master/CONTRIBUTING.md 10 | -------------------------------------------------------------------------------- /codecov.yml: -------------------------------------------------------------------------------- 1 | comment: false 2 | -------------------------------------------------------------------------------- /data/pbcTest.rda: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/data/pbcTest.rda -------------------------------------------------------------------------------- /data/pbcTrain.rda: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/data/pbcTrain.rda -------------------------------------------------------------------------------- /docs/CONTRIBUTING.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | NA • survxai 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 33 | 34 | 35 | 36 | 37 | 38 |
39 |
40 | 96 | 97 | 98 |
99 | 100 |
101 |
102 |
103 |

NA

104 |
105 | 106 |
107 |

108 | How to contribute to survxai

109 |
110 |

111 | Did you find a bug? 112 |

113 |

Please follow these rules when reporting bugs:

114 |
    115 |

  • Install to the latest version of survxai from GitHub and check whether the problem still occurs.

    • 116 |

  • Check, if the bug was not already reported by searching Issues.

    • 117 |

  • If you’re unable to find an open issue addressing the problem, open a new one. Be sure to include a title and clear description, and a sample code demonstrating the problem.

    • 118 |
119 |
120 |
121 |

122 | Did you fixed a bug? 123 |

124 |
    125 |

  • Ensure that changes meet the requirements of The tidyverse style guide and relevant tests are added.

    • 126 |

  • Open a new GitHub pull request with the solution.

    • 127 |

  • Ensure the PR clearly describes the problem and solution. Include the relevant issue number if applicable.

    • 128 |
129 |
130 |
131 |

132 | Do you intend to add a new feature or change an existing one? 133 |

134 |
    135 |
  • Suggest your change in the Issues and start writing code.
    • 136 |
137 |

Thanks!

138 |

MI2 team

139 |
140 |
141 | 142 |
143 | 144 |
145 | 146 | 147 | 157 |
158 | 159 | 160 | 161 | 162 | -------------------------------------------------------------------------------- /docs/MI2logo.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/MI2logo.jpg -------------------------------------------------------------------------------- /docs/articles/Global_explanations_files/figure-html/unnamed-chunk-3-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/Global_explanations_files/figure-html/unnamed-chunk-3-1.png -------------------------------------------------------------------------------- /docs/articles/Global_explanations_files/figure-html/unnamed-chunk-5-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/Global_explanations_files/figure-html/unnamed-chunk-5-1.png -------------------------------------------------------------------------------- /docs/articles/Global_explanations_files/figure-html/unnamed-chunk-6-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/Global_explanations_files/figure-html/unnamed-chunk-6-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-10-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-10-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-11-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-11-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-12-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-12-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-13-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-13-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-14-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-14-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-14-2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-14-2.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-14-3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-14-3.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-15-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-15-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-16-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-16-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-16-2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-16-2.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-16-3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-16-3.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-17-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-17-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-17-2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-17-2.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-17-3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-17-3.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-18-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-18-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-18-2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-18-2.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-18-3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-18-3.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-19-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-19-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-19-2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-19-2.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-19-3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-19-3.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-20-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-20-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-20-2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-20-2.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-20-3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-20-3.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-21-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-21-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-21-2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-21-2.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-21-3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-21-3.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-23-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-23-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-23-2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-23-2.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-23-3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-23-3.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-25-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-25-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-25-2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-25-2.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-25-3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-25-3.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-26-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-26-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-7-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-7-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-8-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-8-1.png -------------------------------------------------------------------------------- /docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-9-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-9-1.png -------------------------------------------------------------------------------- /docs/articles/Local_explanations_files/figure-html/unnamed-chunk-2-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/Local_explanations_files/figure-html/unnamed-chunk-2-1.png -------------------------------------------------------------------------------- /docs/articles/Local_explanations_files/figure-html/unnamed-chunk-3-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/Local_explanations_files/figure-html/unnamed-chunk-3-1.png -------------------------------------------------------------------------------- /docs/articles/Local_explanations_files/figure-html/unnamed-chunk-4-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/articles/Local_explanations_files/figure-html/unnamed-chunk-4-1.png -------------------------------------------------------------------------------- /docs/articles/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | Articles • survxai 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 33 | 34 | 35 | 36 | 37 | 38 |
39 |
40 | 96 | 97 | 98 |
99 | 100 |
101 |
102 |
103 |

Articles

104 |
105 | 106 |
107 |

All vignettes

108 |

109 | 110 | 116 |
117 |
118 |
119 | 120 | 130 |
131 | 132 | 133 | 134 | 135 | -------------------------------------------------------------------------------- /docs/authors.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | Authors • survxai 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 33 | 34 | 35 | 36 | 37 | 38 |
39 |
40 | 96 | 97 | 98 |
99 | 100 |
101 |
102 |
103 |

Authors

104 |
105 |
    106 |
  • 107 |

    Aleksandra Grudziaz. Author, maintainer. 108 |

    109 |
    • 110 |
  • 111 |

    Alicja Gosiewska. Author. 112 |

    113 |
    • 114 |
  • 115 |

    Przemyslaw Biecek. Author, thesis advisor. 116 |

    117 |
    • 118 |
119 | 120 |
121 | 122 |
123 | 124 | 125 | 135 |
136 | 137 | 138 | 139 | 140 | -------------------------------------------------------------------------------- /docs/docsearch.js: -------------------------------------------------------------------------------- 1 | $(function() { 2 | 3 | // register a handler to move the focus to the search bar 4 | // upon pressing shift + "/" (i.e. "?") 5 | $(document).on('keydown', function(e) { 6 | if (e.shiftKey && e.keyCode == 191) { 7 | e.preventDefault(); 8 | $("#search-input").focus(); 9 | } 10 | }); 11 | 12 | $(document).ready(function() { 13 | // do keyword highlighting 14 | /* modified from https://jsfiddle.net/julmot/bL6bb5oo/ */ 15 | var mark = function() { 16 | 17 | var referrer = document.URL ; 18 | var paramKey = "q" ; 19 | 20 | if (referrer.indexOf("?") !== -1) { 21 | var qs = referrer.substr(referrer.indexOf('?') + 1); 22 | var qs_noanchor = qs.split('#')[0]; 23 | var qsa = qs_noanchor.split('&'); 24 | var keyword = ""; 25 | 26 | for (var i = 0; i < qsa.length; i++) { 27 | var currentParam = qsa[i].split('='); 28 | 29 | if (currentParam.length !== 2) { 30 | continue; 31 | } 32 | 33 | if (currentParam[0] == paramKey) { 34 | keyword = decodeURIComponent(currentParam[1].replace(/\+/g, "%20")); 35 | } 36 | } 37 | 38 | if (keyword !== "") { 39 | $(".contents").unmark({ 40 | done: function() { 41 | $(".contents").mark(keyword); 42 | } 43 | }); 44 | } 45 | } 46 | }; 47 | 48 | mark(); 49 | }); 50 | }); 51 | 52 | /* Search term highlighting ------------------------------*/ 53 | 54 | function matchedWords(hit) { 55 | var words = []; 56 | 57 | var hierarchy = hit._highlightResult.hierarchy; 58 | // loop to fetch from lvl0, lvl1, etc. 59 | for (var idx in hierarchy) { 60 | words = words.concat(hierarchy[idx].matchedWords); 61 | } 62 | 63 | var content = hit._highlightResult.content; 64 | if (content) { 65 | words = words.concat(content.matchedWords); 66 | } 67 | 68 | // return unique words 69 | var words_uniq = [...new Set(words)]; 70 | return words_uniq; 71 | } 72 | 73 | function updateHitURL(hit) { 74 | 75 | var words = matchedWords(hit); 76 | var url = ""; 77 | 78 | if (hit.anchor) { 79 | url = hit.url_without_anchor + '?q=' + escape(words.join(" ")) + '#' + hit.anchor; 80 | } else { 81 | url = hit.url + '?q=' + escape(words.join(" ")); 82 | } 83 | 84 | return url; 85 | } 86 | -------------------------------------------------------------------------------- /docs/jquery.sticky-kit.min.js: -------------------------------------------------------------------------------- 1 | /* 2 | Sticky-kit v1.1.2 | WTFPL | Leaf Corcoran 2015 | http://leafo.net 3 | */ 4 | (function(){var b,f;b=this.jQuery||window.jQuery;f=b(window);b.fn.stick_in_parent=function(d){var A,w,J,n,B,K,p,q,k,E,t;null==d&&(d={});t=d.sticky_class;B=d.inner_scrolling;E=d.recalc_every;k=d.parent;q=d.offset_top;p=d.spacer;w=d.bottoming;null==q&&(q=0);null==k&&(k=void 0);null==B&&(B=!0);null==t&&(t="is_stuck");A=b(document);null==w&&(w=!0);J=function(a,d,n,C,F,u,r,G){var v,H,m,D,I,c,g,x,y,z,h,l;if(!a.data("sticky_kit")){a.data("sticky_kit",!0);I=A.height();g=a.parent();null!=k&&(g=g.closest(k)); 5 | if(!g.length)throw"failed to find stick parent";v=m=!1;(h=null!=p?p&&a.closest(p):b("
"))&&h.css("position",a.css("position"));x=function(){var c,f,e;if(!G&&(I=A.height(),c=parseInt(g.css("border-top-width"),10),f=parseInt(g.css("padding-top"),10),d=parseInt(g.css("padding-bottom"),10),n=g.offset().top+c+f,C=g.height(),m&&(v=m=!1,null==p&&(a.insertAfter(h),h.detach()),a.css({position:"",top:"",width:"",bottom:""}).removeClass(t),e=!0),F=a.offset().top-(parseInt(a.css("margin-top"),10)||0)-q, 6 | u=a.outerHeight(!0),r=a.css("float"),h&&h.css({width:a.outerWidth(!0),height:u,display:a.css("display"),"vertical-align":a.css("vertical-align"),"float":r}),e))return l()};x();if(u!==C)return D=void 0,c=q,z=E,l=function(){var b,l,e,k;if(!G&&(e=!1,null!=z&&(--z,0>=z&&(z=E,x(),e=!0)),e||A.height()===I||x(),e=f.scrollTop(),null!=D&&(l=e-D),D=e,m?(w&&(k=e+u+c>C+n,v&&!k&&(v=!1,a.css({position:"fixed",bottom:"",top:c}).trigger("sticky_kit:unbottom"))),eb&&!v&&(c-=l,c=Math.max(b-u,c),c=Math.min(q,c),m&&a.css({top:c+"px"})))):e>F&&(m=!0,b={position:"fixed",top:c},b.width="border-box"===a.css("box-sizing")?a.outerWidth()+"px":a.width()+"px",a.css(b).addClass(t),null==p&&(a.after(h),"left"!==r&&"right"!==r||h.append(a)),a.trigger("sticky_kit:stick")),m&&w&&(null==k&&(k=e+u+c>C+n),!v&&k)))return v=!0,"static"===g.css("position")&&g.css({position:"relative"}), 8 | a.css({position:"absolute",bottom:d,top:"auto"}).trigger("sticky_kit:bottom")},y=function(){x();return l()},H=function(){G=!0;f.off("touchmove",l);f.off("scroll",l);f.off("resize",y);b(document.body).off("sticky_kit:recalc",y);a.off("sticky_kit:detach",H);a.removeData("sticky_kit");a.css({position:"",bottom:"",top:"",width:""});g.position("position","");if(m)return null==p&&("left"!==r&&"right"!==r||a.insertAfter(h),h.remove()),a.removeClass(t)},f.on("touchmove",l),f.on("scroll",l),f.on("resize", 9 | y),b(document.body).on("sticky_kit:recalc",y),a.on("sticky_kit:detach",H),setTimeout(l,0)}};n=0;for(K=this.length;n 2 | 3 | 5 | 8 | 12 | 13 | -------------------------------------------------------------------------------- /docs/news/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | Changelog • survxai 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 33 | 34 | 35 | 36 | 37 | 38 |
39 |
40 | 96 | 97 | 98 |
99 | 100 |
101 | 102 |
103 |
104 |

Change log

105 |
106 | 107 |
108 |
109 |

110 | survxai 0.2.0

111 |
    112 |
  • CONTRIBUTING.md have been added
    • 113 |
  • Cheatsheet was created
    • 114 |
  • Colour palette for broken_prediction() and ceteris_paribus() was added
    • 115 |
  • New optional arguments for plot.broken_prediction() and plot.ceteris_paribus() was added
    • 116 |
117 |
118 |
119 |

120 | survxai 0.1.0

121 | 127 |
128 |
129 |
130 | 131 | 140 | 141 |
142 | 143 |
144 |
145 |

Developed by Aleksandra Grudziaz, Alicja Gosiewska, Przemyslaw Biecek.

146 |
147 | 148 |
149 |

Site built with pkgdown.

150 |
151 | 152 |
153 |
154 | 155 | 156 | 157 | 158 | -------------------------------------------------------------------------------- /docs/pkgdown.css: -------------------------------------------------------------------------------- 1 | /* Sticker footer */ 2 | body > .container { 3 | display: flex; 4 | padding-top: 60px; 5 | min-height: calc(100vh); 6 | flex-direction: column; 7 | } 8 | 9 | body > .container .row { 10 | flex: 1; 11 | } 12 | 13 | footer { 14 | margin-top: 45px; 15 | padding: 35px 0 36px; 16 | border-top: 1px solid #e5e5e5; 17 | color: #666; 18 | display: flex; 19 | } 20 | footer p { 21 | margin-bottom: 0; 22 | } 23 | footer div { 24 | flex: 1; 25 | } 26 | footer .pkgdown { 27 | text-align: right; 28 | } 29 | footer p { 30 | margin-bottom: 0; 31 | } 32 | 33 | img.icon { 34 | float: right; 35 | } 36 | 37 | img { 38 | max-width: 100%; 39 | } 40 | 41 | /* Section anchors ---------------------------------*/ 42 | 43 | a.anchor { 44 | margin-left: -30px; 45 | display:inline-block; 46 | width: 30px; 47 | height: 30px; 48 | visibility: hidden; 49 | 50 | background-image: url(./link.svg); 51 | background-repeat: no-repeat; 52 | background-size: 20px 20px; 53 | background-position: center center; 54 | } 55 | 56 | .hasAnchor:hover a.anchor { 57 | visibility: visible; 58 | } 59 | 60 | @media (max-width: 767px) { 61 | .hasAnchor:hover a.anchor { 62 | visibility: hidden; 63 | } 64 | } 65 | 66 | 67 | /* Fixes for fixed navbar --------------------------*/ 68 | 69 | .contents h1, .contents h2, .contents h3, .contents h4 { 70 | padding-top: 60px; 71 | margin-top: -60px; 72 | } 73 | 74 | /* Static header placement on mobile devices */ 75 | @media (max-width: 767px) { 76 | .navbar-fixed-top { 77 | position: absolute; 78 | } 79 | .navbar { 80 | padding: 0; 81 | } 82 | } 83 | 84 | 85 | /* Sidebar --------------------------*/ 86 | 87 | #sidebar { 88 | margin-top: 30px; 89 | } 90 | #sidebar h2 { 91 | font-size: 1.5em; 92 | margin-top: 1em; 93 | } 94 | 95 | #sidebar h2:first-child { 96 | margin-top: 0; 97 | } 98 | 99 | #sidebar .list-unstyled li { 100 | margin-bottom: 0.5em; 101 | } 102 | 103 | /* Reference index & topics ----------------------------------------------- */ 104 | 105 | .ref-index th {font-weight: normal;} 106 | .ref-index h2 {font-size: 20px;} 107 | 108 | .ref-index td {vertical-align: top;} 109 | .ref-index .alias {width: 40%;} 110 | .ref-index .title {width: 60%;} 111 | 112 | .ref-index .alias {width: 40%;} 113 | .ref-index .title {width: 60%;} 114 | 115 | .ref-arguments th {text-align: right; padding-right: 10px;} 116 | .ref-arguments th, .ref-arguments td {vertical-align: top;} 117 | .ref-arguments .name {width: 20%;} 118 | .ref-arguments .desc {width: 80%;} 119 | 120 | /* Nice scrolling for wide elements --------------------------------------- */ 121 | 122 | table { 123 | display: block; 124 | overflow: auto; 125 | } 126 | 127 | /* Syntax highlighting ---------------------------------------------------- */ 128 | 129 | pre { 130 | word-wrap: normal; 131 | word-break: normal; 132 | border: 1px solid #eee; 133 | } 134 | 135 | pre, code { 136 | background-color: #f8f8f8; 137 | color: #333; 138 | } 139 | 140 | pre .img { 141 | margin: 5px 0; 142 | } 143 | 144 | pre .img img { 145 | background-color: #fff; 146 | display: block; 147 | height: auto; 148 | } 149 | 150 | code a, pre a { 151 | color: #375f84; 152 | } 153 | table { 154 | display: block; 155 | overflow: auto; 156 | width: 100% !important; 157 | } 158 | 159 | .fl {color: #1514b5;} 160 | .fu {color: #000000;} /* function */ 161 | .ch,.st {color: #036a07;} /* string */ 162 | .kw {color: #264D66;} /* keyword */ 163 | .co {color: #888888;} /* comment */ 164 | 165 | .message { color: black; font-weight: bolder;} 166 | .error { color: orange; font-weight: bolder;} 167 | .warning { color: #6A0366; font-weight: bolder;} 168 | 169 | .navbar-mi2logo { 170 | float: left; 171 | margin-right: 15px; 172 | margin-top: 2px; 173 | } 174 | .navbar-mi2 { 175 | background-color: #4a3c89; 176 | color: #fff !important; 177 | margin-right: 0px; 178 | } 179 | .navbar-mi2 > li > a { 180 | color: #fff !important; 181 | } 182 | .navbar-mi2 > .active > a{ 183 | background-color: #370f54 !important; 184 | } 185 | .navbar-mi2 > .open > a:focus, .nav-pills> .open > a:focus{ 186 | background-color: #370f54 !important; 187 | } 188 | .dropdown-menu > .active > a, .dropdown-menu > .active > a:focus{ 189 | background-color: #370f54 !important; 190 | } 191 | 192 | .contents-mi2 > li > a:focus, .nav-pills > li > a:focus { 193 | background-color: #4a3c89 !important; 194 | color: #fff; 195 | } 196 | .contents-mi2 > li.active > a, .nav-pills > li.active > a{ 197 | background-color: #370f54 !important; 198 | } 199 | .contents-mi2 > li > a, .nav-pills > li > a{ 200 | background-color: #4a3c89 !important; 201 | color: #fff; 202 | } 203 | 204 | .sidebar-logo { 205 | display:block; 206 | margin-left:auto; 207 | margin-right:auto; 208 | text-align: justify; 209 | } -------------------------------------------------------------------------------- /docs/pkgdown.js: -------------------------------------------------------------------------------- 1 | $(function() { 2 | $("#sidebar").stick_in_parent({offset_top: 40}); 3 | $('body').scrollspy({ 4 | target: '#sidebar', 5 | offset: 60 6 | }); 7 | 8 | var cur_path = location.href; 9 | $("#navbar ul li a").each(function(index, value) { 10 | if (value.text == "Home") 11 | return; 12 | if (value.getAttribute("href") === "#") 13 | return; 14 | 15 | var path = value.href; 16 | if (cur_path == path) { 17 | // Add class to parent
  • , and enclosing
  • if in dropdown 18 | var menu_anchor = $(value); 19 | menu_anchor.parent().addClass("active"); 20 | menu_anchor.closest("li.dropdown").addClass("active"); 21 | } 22 | }); 23 | }); 24 | -------------------------------------------------------------------------------- /docs/pkgdown.yml: -------------------------------------------------------------------------------- 1 | pandoc: 1.19.2.1 2 | pkgdown: 1.1.0 3 | pkgdown_sha: ~ 4 | articles: 5 | Custom_predict_for_survival_models: ../../../Github/survxai/vignettes/Custom_predict_for_survival_models.html 6 | Global_explanations: ../../../Github/survxai/vignettes/Global_explanations.html 7 | How_to_compare_models_with_survxai: ../../../Github/survxai/vignettes/How_to_compare_models_with_survxai.html 8 | Local_explanations: ../../../Github/survxai/vignettes/Local_explanations.html 9 | 10 | -------------------------------------------------------------------------------- /docs/reference/pbcTest.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | pbcTest — pbcTest • survxai 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 33 | 34 | 35 | 36 | 37 | 38 |
    39 |
    40 | 96 | 97 | 98 |
    99 | 100 |
    101 |
    102 |
    103 |

    pbcTest

    104 |
    105 | 106 | 107 |

    PBC test set 108 | Data set based on pbc from randomForestSRC package. 109 | The data consists of 138 randomly chosen observations The pbcTest contains only complete cases for each observation. 110 | It contains 5 variables: `status`, `sex`, `bili`, `stage`, and `years`.

    111 | 112 | 113 | 114 |

    Source

    115 | 116 |

    randomForestSRC

    117 | 118 |

    References

    119 | 120 |

    Flemming T.R and Harrington D.P., (1991) Counting Processes and Survival Analysis. New York: Wiley.

    121 | 122 | 123 |

    Examples

    124 | 
    data("pbcTest", package = "survxai")
125 | head(pbcTest)
    #>   status sex bili stage     years
126 | #> 2      0   1  1.1     3 12.328767
127 | #> 3      1   0  1.4     4  2.772603
128 | #> 4      1   1  1.8     4  5.273973
129 | #> 5      0   1  3.4     3  4.120548
130 | #> 8      1   1  0.3     3  6.756164
131 | #> 9      1   1  3.2     2  6.575342
    132 |
    133 | 145 |
    146 | 147 |
    148 |
    149 |

    Developed by Aleksandra Grudziaz, Alicja Gosiewska, Przemyslaw Biecek.

    150 |
    151 | 152 |
    153 |

    Site built with pkgdown.

    154 |
    155 | 156 |
    157 |
    158 | 159 | 160 | 161 | 162 | -------------------------------------------------------------------------------- /docs/reference/pbcTrain.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | pbcTrain — pbcTrain • survxai 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 33 | 34 | 35 | 36 | 37 | 38 |
    39 |
    40 | 96 | 97 | 98 |
    99 | 100 |
    101 |
    102 |
    103 |

    pbcTrain

    104 |
    105 | 106 | 107 |

    PBC train set 108 | Data set based on pbc from randomForestSRC package. 109 | The data consists of 138 randomly chosen observations The pbcTrain contains only complete cases for each observation. 110 | It contains 5 variables: `status`, `sex`, `bili`, `stage`, and `years`.

    111 | 112 | 113 | 114 |

    Source

    115 | 116 |

    randomForestSRC

    117 | 118 |

    References

    119 | 120 |

    Flemming T.R and Harrington D.P., (1991) Counting Processes and Survival Analysis. New York: Wiley.

    121 | 122 | 123 |

    Examples

    124 | 
    data("pbcTrain", package = "survxai")
125 | head(pbcTrain)
    #>     status sex bili stage    years
126 | #> 90       1   0  1.6     2 7.367123
127 | #> 250      0   1  0.6     4 4.846575
128 | #> 130      1   1 17.4     3 3.871233
129 | #> 277      0   1  1.0     3 3.928767
130 | #> 291      0   1  3.2     4 2.468493
131 | #> 15       1   1  0.8     3 9.819178
    132 |
    133 | 145 |
    146 | 147 |
    148 |
    149 |

    Developed by Aleksandra Grudziaz, Alicja Gosiewska, Przemyslaw Biecek.

    150 |
    151 | 152 |
    153 |

    Site built with pkgdown.

    154 |
    155 | 156 |
    157 |
    158 | 159 | 160 | 161 | 162 | -------------------------------------------------------------------------------- /docs/reference/plot.surv_ceteris_paribus_explainer-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/reference/plot.surv_ceteris_paribus_explainer-1.png -------------------------------------------------------------------------------- /docs/reference/plot.surv_explainer-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/reference/plot.surv_explainer-1.png -------------------------------------------------------------------------------- /docs/reference/plot.surv_model_performance_explainer-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/reference/plot.surv_model_performance_explainer-1.png -------------------------------------------------------------------------------- /docs/reference/plot.surv_prediction_breakdown_explainer-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/reference/plot.surv_prediction_breakdown_explainer-1.png -------------------------------------------------------------------------------- /docs/reference/plot.surv_variable_response_explainer-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/docs/reference/plot.surv_variable_response_explainer-1.png -------------------------------------------------------------------------------- /docs/reference/print.surv_ceteris_paribus_explainer.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | Ceteris Paribus Print — print.surv_ceteris_paribus_explainer • survxai 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 33 | 34 | 35 | 36 | 37 | 38 |
    39 |
    40 | 96 | 97 | 98 |
    99 | 100 |
    101 |
    102 |
    103 |

    Ceteris Paribus Print

    104 |
    105 | 106 | 107 |

    Ceteris Paribus Print

    108 | 109 | 110 | 
    # S3 method for surv_ceteris_paribus_explainer
111 | print(x, ...)
    112 | 113 |

    Arguments

    114 | 115 | 116 | 117 | 118 | 119 | 120 | 121 | 122 | 123 | 124 |
    x

    the model of 'surv_ceteris_paribus_explainer' class

    ...

    other parameters

    125 | 126 |

    Value

    127 | 128 |

    a data frame

    129 | 130 | 131 |
    132 | 141 |
    142 | 143 |
    144 |
    145 |

    Developed by Aleksandra Grudziaz, Alicja Gosiewska, Przemyslaw Biecek.

    146 |
    147 | 148 |
    149 |

    Site built with pkgdown.

    150 |
    151 | 152 |
    153 |
    154 | 155 | 156 | 157 | 158 | -------------------------------------------------------------------------------- /docs/reference/print.surv_explainer.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | Print Survival Explainer Summary — print.surv_explainer • survxai 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 33 | 34 | 35 | 36 | 37 | 38 |
    39 |
    40 | 96 | 97 | 98 |
    99 | 100 |
    101 |
    102 |
    103 |

    Print Survival Explainer Summary

    104 |
    105 | 106 | 107 |

    Print Survival Explainer Summary

    108 | 109 | 110 | 
    # S3 method for surv_explainer
111 | print(x, ...)
    112 | 113 |

    Arguments

    114 | 115 | 116 | 117 | 118 | 119 | 120 | 121 | 122 | 123 | 124 |
    x

    a model survival expaliner created with the `explain()` function

    ...

    other parameters

    125 | 126 | 127 |
    128 | 135 |
    136 | 137 |
    138 |
    139 |

    Developed by Aleksandra Grudziaz, Alicja Gosiewska, Przemyslaw Biecek.

    140 |
    141 | 142 |
    143 |

    Site built with pkgdown.

    144 |
    145 | 146 |
    147 |
    148 | 149 | 150 | 151 | 152 | -------------------------------------------------------------------------------- /docs/reference/print.surv_model_performance_explainer.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | Print Survival Model Performance — print.surv_model_performance_explainer • survxai 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 33 | 34 | 35 | 36 | 37 | 38 |
    39 |
    40 | 96 | 97 | 98 |
    99 | 100 |
    101 |
    102 |
    103 |

    Print Survival Model Performance

    104 |
    105 | 106 | 107 |

    Print Survival Model Performance

    108 | 109 | 110 | 
    # S3 method for surv_model_performance_explainer
111 | print(x, times = NULL, ...)
    112 | 113 |

    Arguments

    114 | 115 | 116 | 117 | 118 | 119 | 120 | 121 | 122 | 123 | 124 | 125 | 126 | 127 | 128 |
    x

    a model to be explained, object of the class 'model_performance_explainer'

    times

    a vector of integer times on which we want to check the value of prediction error

    ...

    other parameters

    129 | 130 | 131 |
    132 | 139 |
    140 | 141 |
    142 |
    143 |

    Developed by Aleksandra Grudziaz, Alicja Gosiewska, Przemyslaw Biecek.

    144 |
    145 | 146 |
    147 |

    Site built with pkgdown.

    148 |
    149 | 150 |
    151 |
    152 | 153 | 154 | 155 | 156 | -------------------------------------------------------------------------------- /docs/reference/print.surv_prediction_breakdown_explainer.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | Prediction Breakdown Print — print.surv_prediction_breakdown_explainer • survxai 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 33 | 34 | 35 | 36 | 37 | 38 |
    39 |
    40 | 96 | 97 | 98 |
    99 | 100 |
    101 |
    102 |
    103 |

    Prediction Breakdown Print

    104 |
    105 | 106 | 107 |

    Prediction Breakdown Print

    108 | 109 | 110 | 
    # S3 method for surv_prediction_breakdown_explainer
111 | print(x, ..., digits = 3,
112 |   rounding_function = round)
    113 | 114 |

    Arguments

    115 | 116 | 117 | 118 | 119 | 120 | 121 | 122 | 123 | 124 | 125 | 126 | 127 | 129 | 130 | 131 | 132 | 135 | 136 |
    x

    the model model of 'surv_prediction_breakdown_explainer' class

    ...

    other parameters

    digits

    number of decimal places (round) or significant digits (signif) to be used 128 | See the rounding_function argument

    rounding_function

    function that is to used for rounding numbers. 133 | It may be signif() which keeps a specified number of significant digits. 134 | Or the default round() to have the same precision for all components

    137 | 138 | 139 |
    140 | 147 |
    148 | 149 |
    150 |
    151 |

    Developed by Aleksandra Grudziaz, Alicja Gosiewska, Przemyslaw Biecek.

    152 |
    153 | 154 |
    155 |

    Site built with pkgdown.

    156 |
    157 | 158 |
    159 |
    160 | 161 | 162 | 163 | 164 | -------------------------------------------------------------------------------- /docs/reference/print.surv_variable_response_explainer.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | Variable Response Print — print.surv_variable_response_explainer • survxai 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 33 | 34 | 35 | 36 | 37 | 38 |
    39 |
    40 | 96 | 97 | 98 |
    99 | 100 |
    101 |
    102 |
    103 |

    Variable Response Print

    104 |
    105 | 106 | 107 |

    Variable Response Print

    108 | 109 | 110 | 
    # S3 method for surv_variable_response_explainer
111 | print(x, ...)
    112 | 113 |

    Arguments

    114 | 115 | 116 | 117 | 118 | 119 | 120 | 121 | 122 | 123 | 124 |
    x

    the model of 'surv_variable_response_explainer' class

    ...

    other parameters

    125 | 126 |

    Value

    127 | 128 |

    a data frame

    129 | 130 | 131 |
    132 | 141 |
    142 | 143 |
    144 |
    145 |

    Developed by Aleksandra Grudziaz, Alicja Gosiewska, Przemyslaw Biecek.

    146 |
    147 | 148 |
    149 |

    Site built with pkgdown.

    150 |
    151 | 152 |
    153 |
    154 | 155 | 156 | 157 | 158 | -------------------------------------------------------------------------------- /docs/reference/theme_mi2.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | MI^2 plot theme — theme_mi2 • survxai 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 33 | 34 | 35 | 36 | 37 | 38 |
    39 |
    40 | 96 | 97 | 98 |
    99 | 100 |
    101 |
    102 |
    103 |

    MI^2 plot theme

    104 |
    105 | 106 | 107 |

    ggplot theme for charts generated with MI^2 Data Lab packages.

    108 | 109 | 110 | 
    theme_mi2()
    111 | 112 |

    Value

    113 | 114 |

    theme object that can be added to ggplot2 plots

    115 | 116 | 117 |
    118 | 126 |
    127 | 128 | 138 |
    139 | 140 | 141 | 142 | 143 | -------------------------------------------------------------------------------- /inst/CITATION: -------------------------------------------------------------------------------- 1 | bibentry(bibtype = "Article", 2 | author = c( 3 | person(given = "Aleksandra", "Grudziaz"), 4 | person(given = "Alicja", family = "Gosiewska"), 5 | person(given = "Przemyslaw", family = "Biecek") 6 | ), 7 | title = "{survxai}: an {R} package for structure-agnostic explanations of survival models", 8 | volume = 3, 9 | journal = "Journal of Open Source Software", 10 | archivePrefix = "arXiv", 11 | eprint = "1809.07763", 12 | primaryClass = "stat.CO", 13 | year = "2018", 14 | url= "http://dx.doi.org/10.21105/joss.00961", 15 | DOI = "10.21105/joss.00961", 16 | number = 31, 17 | publisher = "The Open Journal", 18 | year = 2018, 19 | month = "Nov", 20 | pages = 961 21 | ) 22 | -------------------------------------------------------------------------------- /man/ceteris_paribus.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/ceteris_paribus.R 3 | \name{ceteris_paribus} 4 | \alias{ceteris_paribus} 5 | \title{Ceteris Paribus} 6 | \usage{ 7 | ceteris_paribus( 8 | explainer, 9 | observation, 10 | grid_points = 5, 11 | selected_variables = NULL 12 | ) 13 | } 14 | \arguments{ 15 | \item{explainer}{a model to be explained, preprocessed by the 'survxai::explain' function} 16 | 17 | \item{observation}{a new observation for which predictions need to be explained} 18 | 19 | \item{grid_points}{grid_points number of points used for response path} 20 | 21 | \item{selected_variables}{if specified, then only these variables will be explained} 22 | } 23 | \value{ 24 | An object of the class surv_ceteris_paribus_explainer. 25 | It's a data frame with calculated average responses. 26 | } 27 | \description{ 28 | The \code{ceteris_paribus()} function computes the predictions for the neighbor of our chosen observation. The neighbour is defined as the observations with changed value of one of the variable. 29 | } 30 | \examples{ 31 | \donttest{ 32 | library(survxai) 33 | library(rms) 34 | data("pbcTrain") 35 | data("pbcTest") 36 | predict_times <- function(model, data, times){ 37 | prob <- rms::survest(model, data, times = times)$surv 38 | return(prob) 39 | } 40 | cph_model <- cph(Surv(years, status)~ sex + bili + stage, 41 | data = pbcTrain, surv = TRUE, x = TRUE, y=TRUE) 42 | surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 43 | y = Surv(pbcTest$years, pbcTest$status), 44 | predict_function = predict_times) 45 | cp_cph <- ceteris_paribus(surve_cph, pbcTest[1,-c(1,5)]) 46 | } 47 | } 48 | -------------------------------------------------------------------------------- /man/explain.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/explain.R 3 | \name{explain} 4 | \alias{explain} 5 | \alias{explain.default} 6 | \title{Create Survival Model Explainer} 7 | \usage{ 8 | explain( 9 | model, 10 | data = NULL, 11 | y, 12 | times = NULL, 13 | predict_function = yhat, 14 | link = I, 15 | label = tail(class(model), 1), 16 | ... 17 | ) 18 | 19 | \method{explain}{default}( 20 | model, 21 | data = NULL, 22 | y, 23 | times = NULL, 24 | predict_function = yhat, 25 | link = I, 26 | label = tail(class(model), 1), 27 | ... 28 | ) 29 | } 30 | \arguments{ 31 | \item{model}{object - a survival model to be explained} 32 | 33 | \item{data}{data.frame, tibble or matrix - data that will be used by survival explainers. If not provided then will be extracted from the model} 34 | 35 | \item{y}{object of class 'surv', contains event status and times} 36 | 37 | \item{times}{optional argument, the vector of time points on which survival probability will be predicted} 38 | 39 | \item{predict_function}{function that takes three arguments: model, new data, vector with times, and returns numeric vector or matrix with predictions. If not passed, function \code{\link[pec]{predictSurvProb}} is used.} 40 | 41 | \item{link}{function - a transformation/link function that shall be applied to raw model predictions} 42 | 43 | \item{label}{character - the name of the survival model. By default it's extracted from the 'class' attribute of the model.} 44 | 45 | \item{...}{other parameters} 46 | } 47 | \value{ 48 | An object of the class 'surv_explainer'. 49 | 50 | It's a list with following fields: 51 | 52 | \itemize{ 53 | \item \code{model} the explained model 54 | \item \code{data} the dataset 55 | \item \code{y} event statuses and times 56 | \item \code{times} time points on which survival probability is predicted 57 | \item \code{predict_function} function that may be used for model predictions, shall return a single numerical value for each time. 58 | \item \code{link} function - a transformation/link function that shall be applied to raw model predictions 59 | \item \code{class} class/classes of a model 60 | \item \code{label} label, by default it's the last value from the \code{class} vector, but may be set to any character. 61 | } 62 | } 63 | \description{ 64 | Survival models may have very different structures. 65 | This function creates a unified representation of a survival model, which can be further processed by various survival 66 | explainers (see also \code{\link[DALEX]{explain}}). 67 | 68 | Please NOTE, that the \code{model} is actually the only required argument. 69 | But some survival explainers may require additional arguments. 70 | } 71 | \examples{ 72 | \donttest{ 73 | library(survxai) 74 | library(rms) 75 | library(randomForestSRC) 76 | data(pbc, package = "randomForestSRC") 77 | pbc <- pbc[complete.cases(pbc),] 78 | predict_times <- function(model, data, times){ 79 | prob <- rms::survest(model, data, times = times)$surv 80 | return(prob) 81 | } 82 | cph_model <- cph(Surv(days/365, status)~ sex + bili + stage, data=pbc, surv=TRUE, x = TRUE, y=TRUE) 83 | surve_cph <- explain(model = cph_model, data = pbc[,-c(1,2)], y = Surv(pbc$days/365, pbc$status), 84 | predict_function = predict_times) 85 | } 86 | } 87 | -------------------------------------------------------------------------------- /man/model_performance.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/model_performance.R 3 | \name{model_performance} 4 | \alias{model_performance} 5 | \title{Model performance for survival models} 6 | \usage{ 7 | model_performance(explainer, type = "BS") 8 | } 9 | \arguments{ 10 | \item{explainer}{a model to be explained, preprocessed by the 'survxai::explain' function} 11 | 12 | \item{type}{character - type of the response to be calculated 13 | Currently following options are implemented: 'BS' for Expected Brier Score} 14 | } 15 | \description{ 16 | Function \code{model_performance} calculates the prediction error for chosen survival model. 17 | } 18 | \details{ 19 | For \code{type = "BS"} prediction error is the time dependent estimates of the population average Brier score. 20 | At a given time point t, the Brier score for a single observation is the squared difference between observed survival status 21 | and a model based prediction of surviving time t. 22 | } 23 | \examples{ 24 | \donttest{ 25 | library(survxai) 26 | library(rms) 27 | data("pbcTrain") 28 | data("pbcTest") 29 | cph_model <- cph(Surv(years, status)~ sex + bili + stage, 30 | data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 31 | surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 32 | y = Surv(pbcTest$years, pbcTest$status)) 33 | mp_cph <- model_performance(surve_cph) 34 | } 35 | 36 | } 37 | \references{ 38 | Ulla B. Mogensen, Hemant Ishwaran, Thomas A. Gerds (2012). Evaluating Random Forests for Survival Analysis Using Prediction Error Curves. Journal of Statistical Software, 50(11), 1-23. URL http://www.jstatsoft.org/v50/i11/. 39 | } 40 | -------------------------------------------------------------------------------- /man/pbcTest.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/pbcTest.R 3 | \docType{data} 4 | \name{pbcTest} 5 | \alias{pbcTest} 6 | \title{pbcTest} 7 | \source{ 8 | randomForestSRC 9 | } 10 | \description{ 11 | PBC test set 12 | Data set based on \code{pbc} from \code{randomForestSRC} package. 13 | The data consists of 138 randomly chosen observations The \code{pbcTest} contains only complete cases for each observation. 14 | It contains 5 variables: `status`, `sex`, `bili`, `stage`, and `years`. 15 | } 16 | \examples{ 17 | data("pbcTest", package = "survxai") 18 | head(pbcTest) 19 | } 20 | \references{ 21 | Flemming T.R and Harrington D.P., (1991) Counting Processes and Survival Analysis. New York: Wiley. 22 | } 23 | -------------------------------------------------------------------------------- /man/pbcTrain.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/pbcTrain.R 3 | \docType{data} 4 | \name{pbcTrain} 5 | \alias{pbcTrain} 6 | \title{pbcTrain} 7 | \source{ 8 | randomForestSRC 9 | } 10 | \description{ 11 | PBC train set 12 | Data set based on \code{pbc} from \code{randomForestSRC} package. 13 | The data consists of 138 randomly chosen observations The \code{pbcTrain} contains only complete cases for each observation. 14 | It contains 5 variables: `status`, `sex`, `bili`, `stage`, and `years`. 15 | } 16 | \examples{ 17 | data("pbcTrain", package = "survxai") 18 | head(pbcTrain) 19 | } 20 | \references{ 21 | Flemming T.R and Harrington D.P., (1991) Counting Processes and Survival Analysis. New York: Wiley. 22 | } 23 | -------------------------------------------------------------------------------- /man/plot.surv_ceteris_paribus_explainer.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/plot_ceteris_paribus.R 3 | \name{plot.surv_ceteris_paribus_explainer} 4 | \alias{plot.surv_ceteris_paribus_explainer} 5 | \title{Plot for ceteris_paribus object} 6 | \usage{ 7 | \method{plot}{surv_ceteris_paribus_explainer}( 8 | x, 9 | ..., 10 | selected_variable = NULL, 11 | scale_type = "factor", 12 | scale_col = NULL, 13 | ncol = 1 14 | ) 15 | } 16 | \arguments{ 17 | \item{x}{object of class "surv_ceteris_paribus_explainer"} 18 | 19 | \item{...}{arguments to be passed to methods, such as graphical parameters for function \code{\link[ggplot2]{geom_step}}.} 20 | 21 | \item{selected_variable}{name of variable we want to draw ceteris paribus plot} 22 | 23 | \item{scale_type}{type of scale of colors, either "discrete" or "gradient"} 24 | 25 | \item{scale_col}{vector containing values of low and high ends of the gradient, when "gradient" type of scale was chosen} 26 | 27 | \item{ncol}{number of columns for faceting} 28 | } 29 | \description{ 30 | Function plot for ceteris_paribus object visualise estimated survival curve of mean probabilities in chosen time points. Black lines on each plot correspond to survival curve for our new observation specified in the \code{ceteris_paribus} function. 31 | } 32 | \examples{ 33 | \donttest{ 34 | library(survxai) 35 | library(rms) 36 | data("pbcTest") 37 | data("pbcTrain") 38 | predict_times <- function(model, data, times){ 39 | prob <- rms::survest(model, data, times = times)$surv 40 | return(prob) 41 | } 42 | cph_model <- cph(Surv(years, status)~sex + bili + stage, data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 43 | surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 44 | y = Surv(pbcTest$years, pbcTest$status), predict_function = predict_times) 45 | cp_cph <- ceteris_paribus(surve_cph, pbcTest[1,-c(1,5)]) 46 | plot(cp_cph) 47 | } 48 | } 49 | -------------------------------------------------------------------------------- /man/plot.surv_explainer.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/plot_explainer.R 3 | \name{plot.surv_explainer} 4 | \alias{plot.surv_explainer} 5 | \title{Plot for surv_explainer object} 6 | \usage{ 7 | \method{plot}{surv_explainer}(x, ...) 8 | } 9 | \arguments{ 10 | \item{x}{object of class "surv_explainer"} 11 | 12 | \item{...}{other arguments for function \code{\link[survminer]{ggsurvplot}}} 13 | } 14 | \description{ 15 | Function plot for surv_explainer object visualise estimated survival curve of mean probabilities in chosen time points. 16 | } 17 | \examples{ 18 | \donttest{ 19 | library(survxai) 20 | library(rms) 21 | data("pbcTest") 22 | data("pbcTrain") 23 | predict_times <- function(model, data, times){ 24 | prob <- rms::survest(model, data, times = times)$surv 25 | return(prob) 26 | } 27 | cph_model <- cph(Surv(years, status)~sex + bili + stage, data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 28 | surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 29 | y = Surv(pbcTest$years, pbcTest$status), predict_function = predict_times) 30 | plot(surve_cph) 31 | } 32 | } 33 | -------------------------------------------------------------------------------- /man/plot.surv_model_performance_explainer.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/plot_model_performance.R 3 | \name{plot.surv_model_performance_explainer} 4 | \alias{plot.surv_model_performance_explainer} 5 | \title{Plot for surv_model_performance object} 6 | \usage{ 7 | \method{plot}{surv_model_performance_explainer}(x, ...) 8 | } 9 | \arguments{ 10 | \item{x}{object of class "surv_model_performance"} 11 | 12 | \item{...}{optional, additional objects of class "surv_model_performance_explainer"} 13 | } 14 | \description{ 15 | Function plot for surv_model_performance object. 16 | } 17 | \examples{ 18 | \donttest{ 19 | library(survxai) 20 | library(rms) 21 | data("pbcTest") 22 | data("pbcTrain") 23 | predict_times <- function(model, data, times){ 24 | prob <- rms::survest(model, data, times = times)$surv 25 | return(prob) 26 | } 27 | cph_model <- cph(Surv(years, status)~sex + bili + stage, data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 28 | surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 29 | y = Surv(pbcTest$years, pbcTest$status), predict_function = predict_times) 30 | mp_cph <- model_performance(surve_cph) 31 | plot(mp_cph) 32 | } 33 | 34 | } 35 | -------------------------------------------------------------------------------- /man/plot.surv_prediction_breakdown_explainer.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/plot_prediction_breakdown.R 3 | \name{plot.surv_prediction_breakdown_explainer} 4 | \alias{plot.surv_prediction_breakdown_explainer} 5 | \title{Plot for surv_breakdown object} 6 | \usage{ 7 | \method{plot}{surv_prediction_breakdown_explainer}( 8 | x, 9 | ..., 10 | numerate = TRUE, 11 | lines = TRUE, 12 | lines_type = 1, 13 | lines_col = "black", 14 | scale_col = c("#010059", "#e0f6fb") 15 | ) 16 | } 17 | \arguments{ 18 | \item{x}{an object of class "surv_prediction_breakdown_explainer"} 19 | 20 | \item{...}{optional, additional objects of class "surv_prediction_breakdown_explainer"} 21 | 22 | \item{numerate}{logical; indicating whether we want to number curves} 23 | 24 | \item{lines}{logical; indicating whether we want to add lines on chosen time point or probability} 25 | 26 | \item{lines_type}{a type of line; see http://sape.inf.usi.ch/quick-reference/ggplot2/linetype} 27 | 28 | \item{lines_col}{a color of line} 29 | 30 | \item{scale_col}{a vector containig two colors for gradient scale in legend} 31 | } 32 | \description{ 33 | Function plot for surv_breakdown object visualise estimated survival curve of mean probabilities in chosen time points. 34 | } 35 | \examples{ 36 | \donttest{ 37 | library(survxai) 38 | library(rms) 39 | data("pbcTest") 40 | data("pbcTrain") 41 | predict_times <- function(model, data, times){ 42 | prob <- rms::survest(model, data, times = times)$surv 43 | return(prob) 44 | } 45 | cph_model <- cph(Surv(years, status)~sex + bili + stage, data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 46 | surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 47 | y = Surv(pbcTest$years, pbcTest$status), predict_function = predict_times) 48 | broken_prediction <- prediction_breakdown(surve_cph, pbcTest[1,-c(1,5)]) 49 | plot(broken_prediction) 50 | } 51 | } 52 | -------------------------------------------------------------------------------- /man/plot.surv_variable_response_explainer.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/plot_variable_response.R 3 | \name{plot.surv_variable_response_explainer} 4 | \alias{plot.surv_variable_response_explainer} 5 | \title{Plot for surv_variable_response object} 6 | \usage{ 7 | \method{plot}{surv_variable_response_explainer}(x, ..., split = "model") 8 | } 9 | \arguments{ 10 | \item{x}{an object of class "surv_variable_response"} 11 | 12 | \item{...}{optional, additional objects of class "surv_variable_response_explainer"} 13 | 14 | \item{split}{a character, either "model" or "variable"; sets the variable for faceting} 15 | } 16 | \description{ 17 | Function plot for surv_variable_response object shows the expected output condition on a selected variable. 18 | } 19 | \examples{ 20 | \donttest{ 21 | library(survxai) 22 | library(rms) 23 | data("pbcTest") 24 | data("pbcTrain") 25 | predict_times <- function(model, data, times){ 26 | prob <- rms::survest(model, data, times = times)$surv 27 | return(prob) 28 | } 29 | cph_model <- cph(Surv(years, status)~sex + bili + stage, data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 30 | surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 31 | y = Surv(pbcTest$years, pbcTest$status), predict_function = predict_times) 32 | svr_cph <- variable_response(surve_cph, "sex") 33 | plot(svr_cph) 34 | } 35 | 36 | } 37 | -------------------------------------------------------------------------------- /man/prediction_breakdown.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/prediction_breakdown.R 3 | \name{prediction_breakdown} 4 | \alias{prediction_breakdown} 5 | \title{BreakDown for survival models} 6 | \usage{ 7 | prediction_breakdown(explainer, observation, time = NULL, prob = NULL, ...) 8 | } 9 | \arguments{ 10 | \item{explainer}{an object of the class 'surv_explainer'} 11 | 12 | \item{observation}{a new observation to explain} 13 | 14 | \item{time}{a time point at which variable contributions are computed. If NULL median time is taken.} 15 | 16 | \item{prob}{a survival probability at which variable contributions are computed} 17 | 18 | \item{...}{other parameters corresponding to arguments from \code{\link[breakDown]{broken}} function from \code{breakDown} package. See https://github.com/pbiecek/breakDown/blob/master/R/break_agnostic.R for details} 19 | } 20 | \value{ 21 | An object of class surv_prediction_breakdown_explainer 22 | } 23 | \description{ 24 | Function \code{prediction_breakdown} is an extension of a broken function from breakDown package. It computes the contribution in prediction for the variables in the model. 25 | The contribution is defined as the difference between survival probabilities for model with added specific value of variable and with the random levels of this variable. 26 | } 27 | \examples{ 28 | \donttest{ 29 | library(survxai) 30 | library(rms) 31 | data("pbcTest") 32 | data("pbcTrain") 33 | predict_times <- function(model, data, times){ 34 | prob <- rms::survest(model, data, times = times)$surv 35 | return(prob) 36 | } 37 | cph_model <- cph(Surv(years, status)~sex + bili + stage, data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 38 | surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 39 | y = Surv(pbcTest$years, pbcTest$status), predict_function = predict_times) 40 | broken_prediction <- prediction_breakdown(surve_cph, pbcTest[1,-c(1,5)]) 41 | } 42 | } 43 | -------------------------------------------------------------------------------- /man/print.surv_ceteris_paribus_explainer.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/print_ceteris_paribus.R 3 | \name{print.surv_ceteris_paribus_explainer} 4 | \alias{print.surv_ceteris_paribus_explainer} 5 | \title{Ceteris Paribus Print} 6 | \usage{ 7 | \method{print}{surv_ceteris_paribus_explainer}(x, ...) 8 | } 9 | \arguments{ 10 | \item{x}{the model of 'surv_ceteris_paribus_explainer' class} 11 | 12 | \item{...}{further arguments passed to or from other methods} 13 | } 14 | \value{ 15 | a data frame 16 | } 17 | \description{ 18 | Ceteris Paribus Print 19 | } 20 | -------------------------------------------------------------------------------- /man/print.surv_explainer.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/print_explainer.R 3 | \name{print.surv_explainer} 4 | \alias{print.surv_explainer} 5 | \title{Print Survival Explainer Summary} 6 | \usage{ 7 | \method{print}{surv_explainer}(x, ...) 8 | } 9 | \arguments{ 10 | \item{x}{a model survival expaliner created with the `explain()` function} 11 | 12 | \item{...}{further arguments passed to or from other methods} 13 | } 14 | \description{ 15 | Print Survival Explainer Summary 16 | } 17 | -------------------------------------------------------------------------------- /man/print.surv_model_performance_explainer.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/print_model_performance.R 3 | \name{print.surv_model_performance_explainer} 4 | \alias{print.surv_model_performance_explainer} 5 | \title{Print Survival Model Performance} 6 | \usage{ 7 | \method{print}{surv_model_performance_explainer}(x, times = NULL, ...) 8 | } 9 | \arguments{ 10 | \item{x}{a model to be explained, object of the class 'model_performance_explainer'} 11 | 12 | \item{times}{a vector of integer times on which we want to check the value of prediction error} 13 | 14 | \item{...}{further arguments passed to or from other methods} 15 | } 16 | \description{ 17 | Print Survival Model Performance 18 | } 19 | -------------------------------------------------------------------------------- /man/print.surv_prediction_breakdown_explainer.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/print_prediction_breakdown.R 3 | \name{print.surv_prediction_breakdown_explainer} 4 | \alias{print.surv_prediction_breakdown_explainer} 5 | \title{Prediction Breakdown Print} 6 | \usage{ 7 | \method{print}{surv_prediction_breakdown_explainer}(x, ..., digits = 3, rounding_function = round) 8 | } 9 | \arguments{ 10 | \item{x}{the model model of 'surv_prediction_breakdown_explainer' class} 11 | 12 | \item{...}{further arguments passed to or from other methods} 13 | 14 | \item{digits}{number of decimal places (round) or significant digits (signif) to be used 15 | See the \code{rounding_function} argument} 16 | 17 | \item{rounding_function}{function that is to used for rounding numbers. 18 | It may be \code{signif()} which keeps a specified number of significant digits. 19 | Or the default \code{round()} to have the same precision for all components} 20 | } 21 | \description{ 22 | Prediction Breakdown Print 23 | } 24 | -------------------------------------------------------------------------------- /man/print.surv_variable_response_explainer.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/print_variable_response.R 3 | \name{print.surv_variable_response_explainer} 4 | \alias{print.surv_variable_response_explainer} 5 | \title{Variable Response Print} 6 | \usage{ 7 | \method{print}{surv_variable_response_explainer}(x, ...) 8 | } 9 | \arguments{ 10 | \item{x}{the model of 'surv_variable_response_explainer' class} 11 | 12 | \item{...}{further arguments passed to or from other methods} 13 | } 14 | \value{ 15 | a data frame 16 | } 17 | \description{ 18 | Variable Response Print 19 | } 20 | -------------------------------------------------------------------------------- /man/theme_mi2.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/theme_mi2.R 3 | \name{theme_mi2} 4 | \alias{theme_mi2} 5 | \title{MI^2 plot theme} 6 | \usage{ 7 | theme_mi2() 8 | } 9 | \value{ 10 | theme object that can be added to ggplot2 plots 11 | } 12 | \description{ 13 | ggplot theme for charts generated with MI^2 Data Lab packages. 14 | } 15 | -------------------------------------------------------------------------------- /man/variable_response.Rd: -------------------------------------------------------------------------------- 1 | % Generated by roxygen2: do not edit by hand 2 | % Please edit documentation in R/variable_response.R 3 | \name{variable_response} 4 | \alias{variable_response} 5 | \title{Variable response for survival models} 6 | \usage{ 7 | variable_response(explainer, variable, type = "pdp", link = explainer$link) 8 | } 9 | \arguments{ 10 | \item{explainer}{an object of the class 'surv_explainer'.} 11 | 12 | \item{variable}{a character with variable name.} 13 | 14 | \item{type}{a character - type of the response to be calculated. 15 | Currently following options are implemented: 'pdp' for Partial Dependency.} 16 | 17 | \item{link}{a function - a link function that shall be applied to raw model predictions. This will be inherited from the explainer.} 18 | } 19 | \description{ 20 | Function \code{variable_response} calculates the expected output condition on a selected variable. 21 | } 22 | \examples{ 23 | \donttest{ 24 | library(survxai) 25 | library(rms) 26 | data("pbcTest") 27 | data("pbcTrain") 28 | predict_times <- function(model, data, times){ 29 | prob <- rms::survest(model, data, times = times)$surv 30 | return(prob) 31 | } 32 | cph_model <- cph(Surv(years, status)~sex + bili + stage, data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 33 | surve_cph <- explain(model = cph_model, data = pbcTest[,-c(1,5)], 34 | y = Surv(pbcTest$years, pbcTest$status), predict_function = predict_times) 35 | svr_cph <- variable_response(surve_cph, "sex") 36 | } 37 | } 38 | -------------------------------------------------------------------------------- /materials/survxai-cheatsheet.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/materials/survxai-cheatsheet.pdf -------------------------------------------------------------------------------- /materials/survxai-cheatsheet.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/materials/survxai-cheatsheet.png -------------------------------------------------------------------------------- /misc/img/breakdown.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/misc/img/breakdown.png -------------------------------------------------------------------------------- /misc/img/ceteris_paribus.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/misc/img/ceteris_paribus.png -------------------------------------------------------------------------------- /misc/img/model_performance.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/misc/img/model_performance.png -------------------------------------------------------------------------------- /misc/img/variable_response.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/misc/img/variable_response.png -------------------------------------------------------------------------------- /misc/paper.bib: -------------------------------------------------------------------------------- 1 | @Manual{ceterisParibus, 2 | title = {ceterisParibus: Ceteris Paribus Plots (What-If Plots) for a Single Observation}, 3 | author = {Biecek, Przemysław}, 4 | year = {2018}, 5 | note = {R package version 0.2.1}, 6 | url = {http://CRAN.R-project.org/package=ceterisParibus} 7 | } 8 | 9 | @article{2018arXiv180608915B, 10 | author = {{Biecek}, Przemysław}, 11 | title = "{DALEX: explainers for complex predictive models}", 12 | url = {http://arxiv.org/abs/1806.08915}, 13 | eprint = {1806.08915}, 14 | primaryClass = "stat.ML", 15 | keywords = {Statistics - Machine Learning, Computer Science - Artificial Intelligence, Computer Science - Machine Learning, Statistics - Applications}, 16 | year = 2018, 17 | adsurl = {http://adsabs.harvard.edu/abs/2018arXiv180608915B}, 18 | adsnote = {Provided by the SAO/NASA Astrophysics Data System} 19 | } 20 | 21 | 22 | @book{collett2015modelling, 23 | title={Modelling Survival Data in Medical Research, Third Edition}, 24 | author={Collett, David}, 25 | isbn={9781498731690}, 26 | series={Chapman \& Hall/CRC Texts in Statistical Science}, 27 | url={http://books.google.pl/books?id=Okf7CAAAQBAJ}, 28 | year={2015}, 29 | publisher={CRC Press} 30 | } 31 | 32 | @article{lu2003modeling, 33 | title={Modeling customer lifetime value using survival analysis—an application in the telecommunications industry}, 34 | author={Lu, Junxiang and Park, O}, 35 | journal={Data Mining Techniques}, 36 | pages={120--128}, 37 | year={2003}, 38 | publisher={Citeseer} 39 | } 40 | 41 | @Article{randomForestSRC, 42 | title = {Random survival forests}, 43 | author = {Ishwaran, Hemant and Kogalur, Udaya B. and Blackstone, Eugene H. and Lauer, Michael S.}, 44 | journal = {The Annals of Applied Statistics}, 45 | year = {2008}, 46 | volume = {2}, 47 | number = {3}, 48 | pages = {841--860}, 49 | doi = {doi:10.1214/08-AOAS169}, 50 | url = {http://arXiv.org/abs/0811.1645v1}, 51 | pdf = {http://arxiv.org/pdf/0811.1645}, 52 | } 53 | 54 | @article{ELEUTERI2003855, 55 | title = {A novel neural network-based survival analysis model}, 56 | journal = {Neural Networks}, 57 | volume = {16}, 58 | number = {5}, 59 | pages = {855 - 864}, 60 | year = {2003}, 61 | note = {Advances in Neural Networks Research: IJCNN '03}, 62 | issn = {0893-6080}, 63 | doi = {10.1016/S0893-6080(03)00098-4}, 64 | url = {http://doi.org/10.1016/S0893-6080(03)00098-4}, 65 | author = {Eleuteri, Antonio and Tagliaferri, Roberto and Milano, Leopoldo and De Placido, Sabino and De Laurentiis, Michele}, 66 | keywords = {Survival analysis, Conditioning probability estimation, Neural networks, Bayesian learning, MCMC methods} 67 | } 68 | 69 | @article{RJ-2017-016, 70 | author = {Greenwell, Brandon M.}, 71 | title = {{pdp: An R Package for Constructing Partial Dependence Plots}}, 72 | year = {2017}, 73 | journal = {{The R Journal}}, 74 | url = {http://journal.r-project.org/archive/2017/RJ-2017-016/index.html}, 75 | pages = {421--436}, 76 | volume = {9}, 77 | number = {1} 78 | } 79 | 80 | @Article{BSScore, 81 | title = {Evaluating Random Forests for Survival Analysis Using 82 | Prediction Error Curves}, 83 | author = {Mogensen, Ulla B. and Ishwaran, Hemant and Gerds, Thomas A. 84 | }, 85 | journal = {Journal of Statistical Software}, 86 | year = {2012}, 87 | volume = {50}, 88 | number = {11}, 89 | pages = {1--23}, 90 | url = {http://www.jstatsoft.org/v50/i11/}, 91 | } 92 | 93 | @ARTICLE{2018arXiv180401955S, 94 | author = {Staniak, Mateusz and Biecek, Przemysław}, 95 | title = "{Explanations of model predictions with live and breakDown packages}", 96 | eprint = {1804.01955}, 97 | primaryClass = "stat.ML", 98 | keywords = {Statistics - Machine Learning, Computer Science - Learning, Statistics - Applications}, 99 | year = 2018, 100 | url = {http://arxiv.org/abs/1804.01955}, 101 | adsurl = {http://adsabs.harvard.edu/abs/2018arXiv180401955S}, 102 | adsnote = {Provided by the SAO/NASA Astrophysics Data System} 103 | } 104 | 105 | 106 | @article{Molnar2018, 107 | doi = {10.21105/joss.00786}, 108 | url = {http://doi.org/10.21105/joss.00786}, 109 | year = {2018}, 110 | month = {jun}, 111 | publisher = {The Open Journal}, 112 | volume = {3}, 113 | number = {26}, 114 | pages = {786}, 115 | author = {Molnar, Christoph}, 116 | title = {iml: An R package for Interpretable Machine Learning}, 117 | journal = {Journal of Open Source Software} 118 | } 119 | 120 | @misc{pramit_choudhary_2018_1198885, 121 | author = {Choudhary, Pramit and 122 | Kramer, Aaron and 123 | datascience.com team, contributors}, 124 | title = {{Skater: Model Interpretation Library}}, 125 | month = mar, 126 | year = 2018, 127 | doi = {10.5281/zenodo.1198885}, 128 | url = {http://doi.org/10.5281/zenodo.1198885} 129 | } 130 | 131 | 132 | @InProceedings{pmlr-v70-dempsey17a, 133 | title = {i{S}urvive: An Interpretable, Event-time Prediction Model for m{H}ealth}, 134 | author = {Dempsey, Walter H. and Moreno, Alexander and Scott, Christy K. and Dennis, Michael L. and Gustafson, David H. and Murphy, Susan A. and Rehg, James M.}, 135 | booktitle = {Proceedings of the 34th International Conference on Machine Learning}, 136 | pages = {970--979}, 137 | year = {2017}, 138 | editor = {Doina Precup and Yee Whye Teh}, 139 | volume = {70}, 140 | series = {Proceedings of Machine Learning Research}, 141 | address = {International Convention Centre, Sydney, Australia}, 142 | month = {06--11 Aug}, 143 | publisher = {PMLR}, 144 | pdf = {http://proceedings.mlr.press/v70/dempsey17a/dempsey17a.pdf}, 145 | url = {http://proceedings.mlr.press/v70/dempsey17a.html} 146 | } 147 | 148 | 149 | 150 | @article{doi:10.4155/fmc.11.23, 151 | author = {Johansson, Ulf and Sönströd, Cecilia and Norinder, Ulf and Boström, Henrik}, 152 | title = {Trade-off between accuracy and interpretability for predictive in silico modeling}, 153 | journal = {Future Medicinal Chemistry}, 154 | volume = {3}, 155 | number = {6}, 156 | pages = {647-663}, 157 | year = {2011}, 158 | doi = {10.4155/fmc.11.23}, 159 | note ={PMID: 21554073}, 160 | URL = {http://doi.org/10.4155/fmc.11.23}, 161 | eprint = {http://doi.org/10.4155/fmc.11.23} 162 | } 163 | 164 | 165 | -------------------------------------------------------------------------------- /survxai.Rproj: -------------------------------------------------------------------------------- 1 | Version: 1.0 2 | 3 | RestoreWorkspace: Default 4 | SaveWorkspace: Default 5 | AlwaysSaveHistory: Default 6 | 7 | EnableCodeIndexing: Yes 8 | UseSpacesForTab: Yes 9 | NumSpacesForTab: 2 10 | Encoding: UTF-8 11 | 12 | RnwWeave: Sweave 13 | LaTeX: pdfLaTeX 14 | 15 | AutoAppendNewline: Yes 16 | StripTrailingWhitespace: Yes 17 | 18 | BuildType: Package 19 | PackageUseDevtools: Yes 20 | PackageInstallArgs: --no-multiarch --with-keep.source 21 | -------------------------------------------------------------------------------- /tests/testthat.R: -------------------------------------------------------------------------------- 1 | library(testthat) 2 | library(survxai) 3 | 4 | test_check("survxai") 5 | -------------------------------------------------------------------------------- /tests/testthat/objects_for_tests.R: -------------------------------------------------------------------------------- 1 | library(survxai) 2 | library(rms) 3 | library(randomForestSRC) 4 | library(prodlim) 5 | library(pec) 6 | library(tibble) 7 | 8 | data("pbcTest") 9 | data("pbcTrain") 10 | pbc2 <- as_tibble(pbcTest) 11 | 12 | predict_times <- function(object, newdata, times){ 13 | prob <- rms::survest(object, newdata, times = times)$surv 14 | return(prob) 15 | } 16 | 17 | predict_times_rf<- function(object, newdata, times, ...){ 18 | f <- sapply(newdata, is.integer) 19 | cols <- names(which(f)) 20 | object$xvar[cols] <- lapply(object$xvar[cols], as.integer) 21 | ptemp <- predict(object,newdata=newdata,importance="none")$survival 22 | pos <- prodlim::sindex(jump.times=object$time.interest,eval.times=times) 23 | p <- cbind(1,ptemp)[,pos+1,drop=FALSE] 24 | if (NROW(p) != NROW(newdata) || NCOL(p) != length(times)) 25 | stop(paste("\nPrediction matrix has wrong dimensions:\nRequested newdata x times: ",NROW(newdata)," x ",length(times),"\nProvided prediction matrix: ",NROW(p)," x ",NCOL(p),"\n\n",sep="")) 26 | p 27 | } 28 | 29 | rf_model <- rfsrc(Surv(years, status)~., data = pbcTrain, ntree = 100) 30 | cph_model <- cph(Surv(years, status)~sex + bili+stage, data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 31 | cph_model2 <- cph(Surv(years, status)~sex+bili, data=pbcTrain, surv=TRUE, x = TRUE, y=TRUE) 32 | 33 | cph_model_different_class <- cph_model 34 | class(cph_model_different_class) <- "custom_model" 35 | 36 | 37 | surve_cph <- explain(model = cph_model, 38 | data = pbcTest[,-c(1,5)], y = Surv(pbcTest$years, pbcTest$status), 39 | predict_function = predict_times) 40 | 41 | surve_cph2 <- explain(model = cph_model2, 42 | data = pbcTest[,-c(1,5)], y = Surv(pbcTest$years, pbcTest$status), 43 | predict_function = predict_times, label = "2") 44 | 45 | 46 | predict_cph <- function(object, newdata, times){ 47 | class(object) <- c("cph", "rms","coxph") 48 | p <- predictSurvProb(object, newdata, times) 49 | p 50 | } 51 | 52 | surve_cph_artificial <- explain(model = cph_model_different_class, 53 | data = pbcTest[,-c(1,5)], y = Surv(pbcTest$years, pbcTest$status), 54 | predict_function = predict_cph) 55 | 56 | surve_cph_tbl <- explain(model = cph_model2, 57 | data = pbc2[,-c(1,5)],y = Surv(pbcTest$years, pbcTest$status), 58 | predict_function = predict_times, label = "2") 59 | 60 | surve_rf <- explain(model = rf_model, 61 | data = pbcTest[,-c(1,5)], y = Surv(pbcTest$years, pbcTest$status), 62 | predict_function = predict_times_rf, label = "rf") 63 | 64 | surve_cph_null_data <- explain(model = cph_model, y = Surv(pbcTest$years, pbcTest$status), 65 | predict_function = predict_times) 66 | surve_cph_null_data$data <- NULL 67 | 68 | explainer <- surve_cph 69 | class(explainer) <- "explainer" 70 | 71 | 72 | 73 | broken_prediction <- prediction_breakdown(surve_cph, pbcTest[1,-c(1,5)]) 74 | broken_prediction_prob <- prediction_breakdown(surve_cph, pbcTest[1,-c(1,5)], prob = 0.9) 75 | broken_prediction2 <- prediction_breakdown(surve_cph2, pbcTest[1,-c(1,5)]) 76 | svr_cph <- variable_response(surve_cph, "sex") 77 | svr_cph2 <- variable_response(surve_cph2, "sex") 78 | svr_cph_group <- variable_response(surve_cph, "bili") 79 | cp_cph <- ceteris_paribus(surve_cph, pbcTest[1,-c(1,5)]) 80 | mp_cph <- model_performance(surve_cph) 81 | mp_cph_artificial <- model_performance(surve_cph_artificial) 82 | 83 | mp_rf <- model_performance(surve_rf) 84 | 85 | 86 | plot_var_resp <- plot(svr_cph) 87 | plot_var_resp_levels <- plot(svr_cph, svr_cph2, split = "variable") 88 | plot_var_resp_default <- plot(svr_cph, svr_cph2) 89 | plot_cp <- plot(cp_cph) 90 | plot_mp <- plot(mp_cph) 91 | 92 | -------------------------------------------------------------------------------- /tests/testthat/test_explainer.R: -------------------------------------------------------------------------------- 1 | context("surv_explainer") 2 | 3 | source("objects_for_tests.R") 4 | 5 | test_that("Creating surv_explainer", { 6 | expect_is(surve_cph, "surv_explainer") 7 | expect_is(surve_cph_null_data, "surv_explainer") 8 | expect_is(surve_cph_tbl, "surv_explainer") 9 | }) 10 | -------------------------------------------------------------------------------- /tests/testthat/test_plot_ceteris_paribus.R: -------------------------------------------------------------------------------- 1 | context("plot_ceteris_paribus") 2 | 3 | source("objects_for_tests.R") 4 | 5 | test_that("Output", { 6 | expect_is(plot(cp_cph), "ggplot") 7 | expect_is(plot(cp_cph, selected_variable = "sex"), "ggplot") 8 | }) 9 | 10 | test_that("Wrong input",{ 11 | expect_error(plot(cp_cph, selected_variable = "se")) 12 | }) -------------------------------------------------------------------------------- /tests/testthat/test_plot_explainer.R: -------------------------------------------------------------------------------- 1 | context("plot_explainer") 2 | 3 | source("objects_for_tests.R") 4 | 5 | test_that("Output", { 6 | expect_is(plot(x=surve_cph), "ggsurvplot") 7 | }) 8 | -------------------------------------------------------------------------------- /tests/testthat/test_plot_model_performance.R: -------------------------------------------------------------------------------- 1 | context("plot_model_performance") 2 | 3 | source("objects_for_tests.R") 4 | 5 | test_that("Output", { 6 | expect_is(plot(mp_cph), "gg") 7 | expect_is(plot(mp_cph, model_performance(surve_cph2)), "gg") 8 | expect_is(plot(mp_cph, mp_cph_artificial), "gg") 9 | }) -------------------------------------------------------------------------------- /tests/testthat/test_plot_prediction_breakdown.R: -------------------------------------------------------------------------------- 1 | context("plot_variable_response") 2 | 3 | source("objects_for_tests.R") 4 | 5 | test_that("Output", { 6 | expect_is(plot(broken_prediction), "ggplot") 7 | expect_is(plot(broken_prediction, broken_prediction2), "ggplot") 8 | expect_is(plot(broken_prediction_prob), "gg") 9 | }) 10 | -------------------------------------------------------------------------------- /tests/testthat/test_plot_variable_response.R: -------------------------------------------------------------------------------- 1 | context("plot_variable_response") 2 | 3 | source("objects_for_tests.R") 4 | 5 | test_that("Output", { 6 | expect_is(plot(svr_cph_group), "gg") 7 | expect_is(plot(svr_cph), "gg") 8 | expect_is(plot_var_resp_levels, "ggplot") 9 | expect_is(plot_var_resp_default, "ggplot") 10 | }) 11 | -------------------------------------------------------------------------------- /tests/testthat/test_prints.R: -------------------------------------------------------------------------------- 1 | context("print functions") 2 | 3 | source("objects_for_tests.R") 4 | 5 | test_that("Output explain", { 6 | expect_error(print(surve_cph), NA) 7 | }) 8 | 9 | 10 | test_that("Output prediction_breakdown", { 11 | expect_error(print(broken_prediction), NA) 12 | }) 13 | 14 | test_that("Output model_performance", { 15 | expect_error(print(mp_cph), NA) 16 | }) 17 | 18 | test_that("Output variable_response", { 19 | expect_error(print(svr_cph), NA) 20 | }) 21 | 22 | test_that("Output ceteris_paribus", { 23 | expect_error(print(cp_cph), NA) 24 | }) 25 | -------------------------------------------------------------------------------- /tests/testthat/test_surv_ceteris_paribus.R: -------------------------------------------------------------------------------- 1 | context("surv_ceteris_paribus") 2 | 3 | source("objects_for_tests.R") 4 | 5 | test_that("Creating surv_ceteris_paribus", { 6 | expect_is(cp_cph, "surv_ceteris_paribus_explainer") 7 | expect_is(cp_cph, "data.frame") 8 | expect_is(ceteris_paribus(surve_cph, pbcTest[1,-c(1,5)], selected_variables = "sex"), "data.frame") 9 | }) 10 | 11 | 12 | test_that("Wrong input",{ 13 | expect_error(ceteris_paribus(surve_cph)) 14 | expect_error(ceteris_paribus(pbcTest[1,-c(1,5)])) 15 | expect_error(ceteris_paribus(surve_cph_null_data, pbcTest[1,-c(1,5)])) 16 | }) -------------------------------------------------------------------------------- /tests/testthat/test_surv_model_performance.R: -------------------------------------------------------------------------------- 1 | context("surv_model_performance") 2 | 3 | source("objects_for_tests.R") 4 | 5 | test_that("Creating surv_model_performance", { 6 | expect_is(mp_cph, "surv_model_performance_explainer") 7 | expect_is(mp_cph, "BS") 8 | expect_is(mp_cph_artificial, "surv_model_performance_explainer") 9 | }) 10 | 11 | test_that("Wrong input",{ 12 | expect_error(model_performance(cph_model)) 13 | expect_error(model_performance(surve_cph_null_data)) 14 | }) -------------------------------------------------------------------------------- /tests/testthat/test_surv_prediction_breakdown.R: -------------------------------------------------------------------------------- 1 | context("surv_prediction_breakdown") 2 | 3 | source("objects_for_tests.R") 4 | 5 | test_that("Creating surv_prediction_breakdown", { 6 | expect_is(broken_prediction, "surv_prediction_breakdown_explainer") 7 | expect_is(broken_prediction, "data.frame") 8 | }) 9 | 10 | 11 | test_that("Wrong input",{ 12 | expect_error(prediction_breakdown(surve_cph_null_data)) 13 | expect_error(prediction_breakdown(explainer)) 14 | }) -------------------------------------------------------------------------------- /tests/testthat/test_surv_variable_response.R: -------------------------------------------------------------------------------- 1 | context("surv_variable_response") 2 | 3 | source("objects_for_tests.R") 4 | 5 | test_that("Creating surv_variable_response", { 6 | expect_is(svr_cph, "surv_variable_response_explainer") 7 | expect_is(svr_cph, "pdp") 8 | expect_is(svr_cph, "data.frame") 9 | }) 10 | 11 | 12 | test_that("Wrong input",{ 13 | expect_error(variable_response(surve_cph_null_data)) 14 | expect_error(variable_response(explainer)) 15 | }) 16 | -------------------------------------------------------------------------------- /vignettes/Custom_predict_for_survival_models.Rmd: -------------------------------------------------------------------------------- 1 | --- 2 | title: "Custom predict function for survival models" 3 | author: "Alicja Gosiewska, Aleksandra Grudziąż" 4 | date: "`r Sys.Date()`" 5 | output: 6 | html_document: 7 | toc: true 8 | toc_float: true 9 | number_sections: true 10 | vignette: > 11 | %\VignetteEngine{knitr::knitr} 12 | %\VignetteIndexEntry{Custom predict function for survival models} 13 | %\usepackage[UTF-8]{inputenc} 14 | --- 15 | 16 | ```{r setup, include=FALSE} 17 | knitr::opts_chunk$set(echo = TRUE, 18 | message = FALSE, 19 | warning = FALSE) 20 | ``` 21 | 22 | # Introduction 23 | 24 | This vignette contains example predict functions for survival models. Some functions are already implemented. Therefore, for some models there is no need to specify predict function. 25 | 26 | ```{r dataset} 27 | data(pbc, package = "randomForestSRC") 28 | pbc <- pbc[complete.cases(pbc),] 29 | pbc$sex <- as.factor(pbc$sex) 30 | pbc$stage <- as.factor(pbc$stage) 31 | ``` 32 | 33 | 34 | # Implemented Models 35 | 36 | Currently implemented model classes. Objects listed below don't need specified predict function. 37 | 38 | * `aalen` 39 | * `riskRegression` 40 | * `cox.aalen` 41 | * `cph` 42 | * `coxph` 43 | * `matrix` 44 | * `selectCox` 45 | * `pecCforest` 46 | * `prodlim` 47 | * `psm` 48 | * `survfit` 49 | * `pecRpart` 50 | * `pecCtree` 51 | 52 | ```{r, models} 53 | set.seed(1024) 54 | library(rms) 55 | library(survxai) 56 | cph_model <- cph(Surv(days/365, status) ~ treatment + age + sex + ascites + hepatom + spiders + edema + bili + chol + albumin + copper + alk + sgot + trig + platelet + prothrombin + stage , data = pbc, surv = TRUE, x = TRUE, y=TRUE) 57 | 58 | surve_cph <- explain(model = cph_model, 59 | data = pbc[,-c(1,2)], y = Surv(pbc$days/365, pbc$status)) 60 | ``` 61 | 62 | # RandomForestSRC 63 | 64 | Predict function for class `rfsrc` is not implemented. Therefore, custom predict function should be provided. 65 | 66 | ```{r} 67 | library(prodlim) 68 | library(randomForestSRC) 69 | 70 | predict_rf <- function(object, newdata, times, ...){ 71 | f <- sapply(newdata, is.integer) 72 | cols <- names(which(f)) 73 | object$xvar[cols] <- lapply(object$xvar[cols], as.integer) 74 | ptemp <- predict(object,newdata=newdata,importance="none")$survival 75 | pos <- prodlim::sindex(jump.times=object$time.interest,eval.times=times) 76 | p <- cbind(1,ptemp)[,pos+1,drop=FALSE] 77 | return(p) 78 | } 79 | ``` 80 | 81 | ```{r} 82 | pbc$year <- pbc$days/365 83 | rf_model <- rfsrc(Surv(year, status)~., data = pbc[,-1]) 84 | 85 | surve_rf <- explain(model = rf_model, 86 | data = pbc[,-c(1,2,20)], y = Surv(pbc$year, pbc$status), 87 | predict_function = predict_rf) 88 | ``` 89 | 90 | 91 | 92 | # survreg 93 | 94 | 95 | Predict function for class `survreg` is not implemented. Therefore, custom predict function should be provided. 96 | 97 | 98 | ```{r} 99 | library(survival) 100 | 101 | predict_reg <- function(model, newdata, times){ 102 | times <- sort(times) 103 | vars <- all.vars(model$call[[2]][[2]]) 104 | n_vars <- which(colnames(newdata) %in% vars) 105 | if(length(n_vars)>0){ 106 | newdata <- newdata[,-c(n_vars)] 107 | } 108 | model$x <- model.matrix(~., newdata) 109 | res <- matrix(ncol = length(times), nrow = nrow(newdata)) 110 | for(i in 1:nrow(newdata)) { 111 | res[i,] <- cfc.survreg.survprob(t = times, args = model, n = i) 112 | } 113 | return(res) 114 | } 115 | 116 | ``` 117 | 118 | ```{r} 119 | reg_model <- survreg(Surv(year, status)~., data = pbc[,-1], x = TRUE) 120 | 121 | surve_reg <- explain(model = rf_model, 122 | data = pbc[,-c(1,2,20)], 123 | y = Surv(pbc$year, pbc$status), 124 | predict_function = predict_reg) 125 | ``` 126 | 127 | 128 | 129 | -------------------------------------------------------------------------------- /vignettes/Global_explanations.Rmd: -------------------------------------------------------------------------------- 1 | --- 2 | title: "Survival models - global explanations" 3 | author: "Alicja Gosiewska" 4 | date: "`r Sys.Date()`" 5 | output: 6 | html_document: 7 | toc: true 8 | toc_float: true 9 | number_sections: true 10 | vignette: > 11 | %\VignetteEngine{knitr::knitr} 12 | %\VignetteIndexEntry{Survival models - global explanations} 13 | %\usepackage[UTF-8]{inputenc} 14 | --- 15 | 16 | ```{r setup, include=FALSE} 17 | knitr::opts_chunk$set(echo = TRUE, 18 | message = FALSE, 19 | warning = FALSE) 20 | ``` 21 | 22 | # Introduction 23 | Package *survxai* contains functions for creating a unified representation of a survival models. Such representations can be further processed by various survival explainers. Tools implemented in *survxai* help to understand how input variables are used in the model and what impact do they have on final model prediction. 24 | 25 | The analyses carried out using this package can be divided into two parts: local analyses of new observations and global analyses showing the structures of survival models. This vignette describes local explanations. 26 | 27 | Methods and functions in *survxai* package are based on [*DALEX* package](https://github.com/pbiecek/DALEX). 28 | 29 | # Use case - data 30 | 31 | ## Data set 32 | In our use case we will use the data from the Mayo Clinic trial in primary biliary cirrhosis (PBC) of the liver conducted between 1974 and 1984. A total of 424 PBC patients, referred to Mayo Clinic during that ten-year interval, met eligibility criteria for the randomized placebo controlled trial of the drug D-penicillamine. The 33 | first 312 cases in the data set participated in the randomized trial and contain largely complete data. 34 | The `pbc` data is included in the [*randomForestSRC* package](https://CRAN.R-project.org/package=randomForestSRC). 35 | ```{r dataset} 36 | data(pbc, package = "randomForestSRC") 37 | pbc <- pbc[complete.cases(pbc),] 38 | 39 | head(pbc) 40 | ``` 41 | 42 | Our original data set contains only the numerical variables. 43 | For this usecase we convert variables `sex` and `stage` to factor variables. 44 | 45 | ```{r} 46 | pbc$sex <- as.factor(pbc$sex) 47 | pbc$stage <- as.factor(pbc$stage) 48 | ``` 49 | 50 | 51 | ## Model 52 | We will create Cox proportional hazards model based on five variables from our data set: `age`, `treatment`, `status`, `sex` and `bili`. 53 | ```{r, models} 54 | set.seed(1024) 55 | library(rms) 56 | library(survxai) 57 | 58 | pbc_smaller <- pbc[,c("days", "status", "treatment", "sex", "age", "bili", "stage")] 59 | pbc_smaller$years <- pbc_smaller$days/356 60 | pbc_smaller <- pbc_smaller[,-1] 61 | head(pbc_smaller) 62 | cph_model <- cph(Surv(years, status) ~ treatment + sex + age + bili + stage, data = pbc_smaller, surv = TRUE, x = TRUE, y=TRUE) 63 | ``` 64 | 65 | # Global explanations 66 | In this section we focus on explanations of the global and conditional model structure. 67 | 68 | ## Explainers 69 | First, we have to create survival explainers - objects to wrap-up the black-box model with meta-data. Explainers unify model interfacing. 70 | 71 | Some models require custom predict function. Examples are in [Explainations of different survival models vignette](https://mi2datalab.github.io/survxai/articles/Custom_predict_for_survival_models.html). 72 | ```{r, explainer} 73 | 74 | surve_cph <- explain(model = cph_model, 75 | data = pbc_smaller[,-c(1,7)], 76 | y = Surv(pbc_smaller$years, pbc_smaller$status)) 77 | print(surve_cph) 78 | ``` 79 | 80 | ## Model performance 81 | Currently, in the *survxai* package is implemented only the `BS` type of model performance. 82 | In this metod for each time point we compute the prediction error for our model. 83 | ```{r} 84 | mp_cph <- model_performance(surve_cph) 85 | print(mp_cph) 86 | ``` 87 | 88 | After creating the `surv_model_prediction` object we can visualize it in a very convinient way using the generic `plot()` function. 89 | On our plot there are prediction error curves for model from the explainer. 90 | For more details about these curves see: [Mogensen, 2012](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4194196/) 91 | 92 | ```{r} 93 | plot(mp_cph) 94 | ``` 95 | 96 | 97 | ## Variable response 98 | Variable response explainers are designed to better understand the relation between a variable and a model output. 99 | These types of explainers are inspired among others by *pdp* package [Greenwell, 2017](https://journal.r-project.org/archive/2017/RJ-2017-016/index.html). 100 | ```{r} 101 | vr_cph_sex <- variable_response(surve_cph, "sex") 102 | print(vr_cph_sex) 103 | vr_cph_bili <- variable_response(surve_cph, "bili") 104 | ``` 105 | 106 | 107 | After creating the `surv_variable_response` objects we can visualize them in a very convinient way using the generic `plot()` function. 108 | 109 | Variable response plots for survival models are survival curves conditioned by one variable. Each curve represent different value of chosen variable. For factor variables curves covers all possible values, for numeric variables values are divided into quantiles. 110 | 111 | Variable response plot illustrates how will the mean survival curve change along with the changing variable value. 112 | ```{r} 113 | plot(vr_cph_sex) 114 | ``` 115 | 116 | ```{r} 117 | plot(vr_cph_bili) 118 | ``` 119 | -------------------------------------------------------------------------------- /vignettes/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-12-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/vignettes/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-12-1.png -------------------------------------------------------------------------------- /vignettes/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-13-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/vignettes/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-13-1.png -------------------------------------------------------------------------------- /vignettes/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-15-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/vignettes/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-15-1.png -------------------------------------------------------------------------------- /vignettes/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-16-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/vignettes/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-16-1.png -------------------------------------------------------------------------------- /vignettes/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-9-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/MI2DataLab/survxai/ee5c7df52b347e422efcf028bf2afe652284fb2d/vignettes/How_to_compare_models_with_survxai_files/figure-html/unnamed-chunk-9-1.png -------------------------------------------------------------------------------- /vignettes/Local_explanations.Rmd: -------------------------------------------------------------------------------- 1 | --- 2 | title: "Survival models - local explanations" 3 | author: "Aleksandra Grudziaz" 4 | date: "`r Sys.Date()`" 5 | output: 6 | html_document: 7 | toc: true 8 | toc_float: true 9 | number_sections: true 10 | vignette: > 11 | %\VignetteEngine{knitr::knitr} 12 | %\VignetteIndexEntry{Survival models - local explanations} 13 | %\usepackage[UTF-8]{inputenc} 14 | --- 15 | 16 | ```{r setup, include=FALSE} 17 | knitr::opts_chunk$set(echo = TRUE, 18 | message = FALSE, 19 | warning = FALSE) 20 | ``` 21 | 22 | # Introduction 23 | Package *survxai* contains functions for creating a unified representation of a survival models. Such representations can be further processed by various survival explainers. Tools implemented in *survxai* help to understand how input variables are used in the model and what impact do they have on final model prediction. 24 | 25 | The analyses carried out using this package can be divided into two parts: local analyses of new observations and global analyses showing the structures of survival models. This vignette describes local explanations. 26 | 27 | Methods and functions in *survxai* package are based on [*DALEX* package](https://github.com/pbiecek/DALEX). 28 | 29 | # Use case - data 30 | 31 | ## Data set 32 | In our use case we will use the data from the Mayo Clinic trial in primary biliary cirrhosis (PBC) of the liver conducted between 1974 and 1984. A total of 424 PBC patients, referred to Mayo Clinic during that ten-year interval, met eligibility criteria for the randomized placebo controlled trial of the drug D-penicillamine. The 33 | first 312 cases in the data set participated in the randomized trial and contain largely complete data. 34 | The `pbc` data is included in the [*randomForestSRC* package](https://CRAN.R-project.org/package=randomForestSRC). 35 | ```{r dataset} 36 | data(pbc, package = "randomForestSRC") 37 | pbc <- pbc[complete.cases(pbc),] 38 | 39 | head(pbc) 40 | ``` 41 | 42 | Our original data set contains only the numerical variables. 43 | For this usecase we convert variables `sex` and `stage` to factor variables. 44 | 45 | ```{r} 46 | pbc$sex <- as.factor(pbc$sex) 47 | pbc$stage <- as.factor(pbc$stage) 48 | ``` 49 | 50 | 51 | ## Models 52 | We will create Cox proportional hazards model based on five variables from our data set: `age`, `treatment`, `status`, `sex` and `bili`. 53 | ```{r, models} 54 | set.seed(1024) 55 | library(rms) 56 | library(survxai) 57 | 58 | pbc_smaller <- pbc[,c("days", "status", "treatment", "sex", "age", "bili", "stage")] 59 | head(pbc_smaller) 60 | 61 | cph_model <- cph(Surv(days/365, status) ~ treatment + sex + age + bili + stage , data = pbc_smaller, surv = TRUE, x = TRUE, y=TRUE) 62 | ``` 63 | 64 | # Local explanations 65 | In this section we will focus on the local explanations - the explanations for chosen new observations. 66 | 67 | ## Explainers 68 | First, we have to create survival explainers - objects to wrap-up the black-box model with meta-data. Explainers unify model interfacing. 69 | 70 | We have to define custom predict function which takes three arguments: model, data and vector with time points. Predict funcions may vary depending on the model. Examples for some models are in [Explainations of different survival models vignette](https://github.com/MI2DataLab/survxai/blob/master/vignettes/Custom_predict_for_survival_models.Rmd). 71 | ```{r, explainer} 72 | 73 | predict_times <- function(model, data, times){ 74 | prob <- rms::survest(model, data, times = times)$surv 75 | return(prob) 76 | } 77 | 78 | surve_cph <- explain(model = cph_model, 79 | data = pbc_smaller[,-c(1,2)], y = Surv(pbc_smaller$days/365, pbc_smaller$status), 80 | predict_function = predict_times) 81 | 82 | print(surve_cph) 83 | ``` 84 | 85 | ## Ceteris paribus 86 | Ceteris Paribus Plots (What-If Plots) are designed to present model responses around a single point in the feature space. 87 | For more details for generalised models of machine learning see: https://github.com/pbiecek/ceterisParibus. 88 | 89 | Ceteris Paribus Plots for survival models are survival curves around one observation. Each curve represent observation with different value of chosen variable. For factor variables curves covers all possible values, for numeric variables values are divided into quantiles. 90 | 91 | Ceteris Paribus Plot illustrates how will the survival curve change along with the changing variable. 92 | 93 | Below, we plot Ceteris Paribus for one observation. 94 | 95 | ```{r, single observation} 96 | single_observation <- pbc_smaller[1,-c(1,2)] 97 | single_observation 98 | ``` 99 | 100 | ```{r, ceteris paribus} 101 | cp_cph <- ceteris_paribus(surve_cph, single_observation) 102 | print(cp_cph) 103 | ``` 104 | 105 | After creating the `surv_ceteris_paribus` object we can visualize it in a very convinient way using the generic `plot()` function. Black line represent prediction of original observation. 106 | 107 | ```{r, fig.height=6} 108 | plot(cp_cph, scale_type = "gradient", scale_col = c("red", "blue"), ncol = 2) 109 | ``` 110 | 111 | We can see that there are differences for stages. Next, we will plot Ceteris Paribus for sigle variable `stage`. 112 | 113 | ```{r, fig.height=3} 114 | plot(cp_cph, selected_variable = "stage", scale_type = "gradient", scale_col = c("red", "blue")) 115 | ``` 116 | 117 | We see a trend that a lower `stage` means a higher probability of survival for chosen observation. 118 | 119 | 120 | ## Prediction breakdown 121 | Break Down Plot presents variable contributions in final predictions. 122 | For more details for generalised models of machine learning see: https://github.com/pbiecek/breakDown. 123 | 124 | Break Down Plots for survival models compare differences in predictions for median of time. 125 | 126 | ```{r, prediction breakdown} 127 | broken_prediction_cph <- prediction_breakdown(surve_cph, pbc_smaller[1,-c(1,2)]) 128 | print(broken_prediction_cph) 129 | ``` 130 | 131 | After creating the `surv_prediction_breakdown` object we can visualize it in a very convinient way using the generic `plot()` function. 132 | 133 | ```{r} 134 | plot(broken_prediction_cph, scale_col = c("red", "blue")) 135 | ``` 136 | 137 | This plot helps to understand the factors that drive survival probability for a single observation. 138 | 139 | 140 | --------------------------------------------------------------------------------