├── ml_app ├── __init__.py ├── modeling │ ├── __init__.py │ ├── ml_inference.py │ └── ml_pipeline.py ├── utils │ ├── __init__.py │ ├── config.py │ └── ml_model_management.py ├── data_processing │ ├── __init__.py │ └── pre_processor.py ├── requirements.in ├── saved_models │ └── census_xgb_artifacts.pkl ├── app.py └── requirements.txt ├── README.md ├── .gitignore ├── notebooks ├── Test ML API.ipynb └── ML Inference Pipeline.ipynb └── LICENSE /ml_app/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /ml_app/modeling/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /ml_app/utils/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /ml_app/data_processing/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /ml_app/requirements.in: -------------------------------------------------------------------------------- 1 | requests 2 | flask 3 | flask_cors 4 | scikit-learn 5 | pandas 6 | numpy 7 | xgboost 8 | dill -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # ml_model_deployment_example 2 | A simple example to showcase machine learning model deployment with an API using Python 3 | -------------------------------------------------------------------------------- /ml_app/saved_models/census_xgb_artifacts.pkl: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dipanjanS/ml_model_deployment_example/HEAD/ml_app/saved_models/census_xgb_artifacts.pkl -------------------------------------------------------------------------------- /ml_app/modeling/ml_inference.py: -------------------------------------------------------------------------------- 1 | 2 | def make_xgb_model_predictions(request_df, ml_model_artifacts): 3 | 4 | # load saved ML model 5 | ml_model = ml_model_artifacts['xgb_model'] 6 | 7 | # make model predictions 8 | predictions = ml_model.predict(request_df) 9 | 10 | # return predictions 11 | return { 12 | 'predicted_classes' : list(predictions) 13 | } -------------------------------------------------------------------------------- /ml_app/utils/config.py: -------------------------------------------------------------------------------- 1 | import os 2 | 3 | PARENT_DIR = os.path.dirname(os.path.dirname(os.path.realpath(__file__))) # get location of parent dir automatically 4 | XGB_ML_ARTIFACT_PATH = PARENT_DIR+'/saved_models/census_xgb_artifacts.pkl' 5 | 6 | 7 | MODEL_HISTORY = { 8 | 'version_1': { 9 | 'model_type' : 'XGBoost', 10 | 'model_artifact_location': XGB_ML_ARTIFACT_PATH 11 | } 12 | } 13 | -------------------------------------------------------------------------------- /ml_app/utils/ml_model_management.py: -------------------------------------------------------------------------------- 1 | import dill 2 | 3 | 4 | def save_model_artifacts(path, ml_artifact): 5 | with open(path, "wb") as dill_outfile: 6 | dill.dump(ml_artifact, dill_outfile) 7 | print('Saved ml artifact at location:', path) 8 | 9 | 10 | def load_model_artifacts(path): 11 | with open(path, "rb") as dill_infile: 12 | ml_artifact = dill.load(dill_infile) 13 | print('Loaded ml artifact from location:', path) 14 | return ml_artifact -------------------------------------------------------------------------------- /ml_app/modeling/ml_pipeline.py: -------------------------------------------------------------------------------- 1 | from ml_app.utils.config import XGB_ML_ARTIFACT_PATH 2 | from ml_app.utils.ml_model_management import load_model_artifacts 3 | from ml_app.data_processing.pre_processor import form_dataset, impute_and_encode_features 4 | from ml_app.modeling.ml_inference import make_xgb_model_predictions 5 | 6 | 7 | 8 | def load_xgb_ml_artifacts(path=XGB_ML_ARTIFACT_PATH): 9 | 10 | # 1. Load model artifacts 11 | ml_artifacts = load_model_artifacts(path=path) 12 | return ml_artifacts 13 | 14 | 15 | 16 | def run_xgb_ml_pipeline(request_data, ml_artifacts): 17 | 18 | # 2. Create request dataset 19 | request_df = form_dataset(request_data=request_data, 20 | ml_model_artifacts=ml_artifacts) 21 | 22 | # 3. Impute and Encode Features 23 | request_df = impute_and_encode_features(request_df=request_df, 24 | ml_model_artifacts=ml_artifacts) 25 | 26 | # 4. Load and make ML model predictions 27 | pred_response = make_xgb_model_predictions(request_df=request_df, 28 | ml_model_artifacts=ml_artifacts) 29 | 30 | # return response 31 | return pred_response -------------------------------------------------------------------------------- /ml_app/app.py: -------------------------------------------------------------------------------- 1 | ## Make module accessible ## 2 | import os.path 3 | import sys 4 | PACKAGE_PARENT = '..' 5 | SCRIPT_DIR = os.path.dirname(os.path.realpath(os.path.join(os.getcwd(), os.path.expanduser(__file__)))) 6 | sys.path.append(os.path.normpath(os.path.join(SCRIPT_DIR, PACKAGE_PARENT))) 7 | ##### 8 | 9 | from flask import Flask, request, jsonify 10 | from flask_cors import CORS # Cross Origin Resource Sharing (CORS), making cross-origin AJAX possible 11 | 12 | from ml_app.modeling import ml_pipeline as mp 13 | 14 | 15 | HEADERS = {'content-type': 'application/json'} 16 | 17 | # Instantiate Flask App 18 | app = Flask(__name__) 19 | CORS(app) 20 | 21 | # This runs as soon as we setup our web service to run 22 | XGB_ML_ARTIFACTS = mp.load_xgb_ml_artifacts() 23 | 24 | 25 | # Liveness test 26 | @app.route('/income_classifier/api/v1/liveness', methods=['GET', 'POST']) 27 | def liveness(): 28 | return 'API Live!' 29 | 30 | 31 | # Model 2 inference endpoint 32 | @app.route('/income_classifier/api/v1/predict', methods=['POST']) 33 | def xgb_model_inference(): 34 | input_data = request.get_json(force=True)['data'] 35 | response = mp.run_xgb_ml_pipeline(input_data, XGB_ML_ARTIFACTS) 36 | return jsonify(response) 37 | 38 | 39 | # running REST interface, port=5000 for direct test 40 | # use debug=True when debugging, NOT when deploying 41 | if __name__ == "__main__": 42 | app.run(debug=False, host='0.0.0.0', port=8900) -------------------------------------------------------------------------------- /ml_app/requirements.txt: -------------------------------------------------------------------------------- 1 | # 2 | # This file is autogenerated by pip-compile 3 | # To update, run: 4 | # 5 | # pip-compile requirements.in 6 | # 7 | certifi==2020.12.5 8 | # via requests 9 | chardet==4.0.0 10 | # via requests 11 | click==7.1.2 12 | # via flask 13 | dill==0.3.3 14 | # via -r requirements.in 15 | flask-cors==3.0.10 16 | # via -r requirements.in 17 | flask==1.1.2 18 | # via 19 | # -r requirements.in 20 | # flask-cors 21 | idna==2.10 22 | # via requests 23 | itsdangerous==1.1.0 24 | # via flask 25 | jinja2==2.11.3 26 | # via flask 27 | joblib==1.0.1 28 | # via scikit-learn 29 | markupsafe==1.1.1 30 | # via jinja2 31 | numpy==1.20.2 32 | # via 33 | # -r requirements.in 34 | # pandas 35 | # scikit-learn 36 | # scipy 37 | # xgboost 38 | pandas==1.2.4 39 | # via -r requirements.in 40 | python-dateutil==2.8.1 41 | # via pandas 42 | pytz==2021.1 43 | # via pandas 44 | requests==2.25.1 45 | # via -r requirements.in 46 | scikit-learn==0.24.1 47 | # via -r requirements.in 48 | scipy==1.6.3 49 | # via 50 | # scikit-learn 51 | # xgboost 52 | six==1.15.0 53 | # via 54 | # flask-cors 55 | # python-dateutil 56 | threadpoolctl==2.1.0 57 | # via scikit-learn 58 | urllib3==1.26.4 59 | # via requests 60 | werkzeug==1.0.1 61 | # via flask 62 | xgboost==1.4.1 63 | # via -r requirements.in 64 | -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- 1 | # Byte-compiled / optimized / DLL files 2 | __pycache__/ 3 | *.py[cod] 4 | *$py.class 5 | 6 | # C extensions 7 | *.so 8 | 9 | # Distribution / packaging 10 | .Python 11 | build/ 12 | develop-eggs/ 13 | dist/ 14 | downloads/ 15 | eggs/ 16 | .eggs/ 17 | lib/ 18 | lib64/ 19 | parts/ 20 | sdist/ 21 | var/ 22 | wheels/ 23 | pip-wheel-metadata/ 24 | share/python-wheels/ 25 | *.egg-info/ 26 | .installed.cfg 27 | *.egg 28 | MANIFEST 29 | 30 | # PyInstaller 31 | # Usually these files are written by a python script from a template 32 | # before PyInstaller builds the exe, so as to inject date/other infos into it. 33 | *.manifest 34 | *.spec 35 | 36 | # Installer logs 37 | pip-log.txt 38 | pip-delete-this-directory.txt 39 | 40 | # Unit test / coverage reports 41 | htmlcov/ 42 | .tox/ 43 | .nox/ 44 | .coverage 45 | .coverage.* 46 | .cache 47 | nosetests.xml 48 | coverage.xml 49 | *.cover 50 | *.py,cover 51 | .hypothesis/ 52 | .pytest_cache/ 53 | 54 | # Translations 55 | *.mo 56 | *.pot 57 | 58 | # Django stuff: 59 | *.log 60 | local_settings.py 61 | db.sqlite3 62 | db.sqlite3-journal 63 | 64 | # Flask stuff: 65 | instance/ 66 | .webassets-cache 67 | 68 | # Scrapy stuff: 69 | .scrapy 70 | 71 | # Sphinx documentation 72 | docs/_build/ 73 | 74 | # PyBuilder 75 | target/ 76 | 77 | # Jupyter Notebook 78 | .ipynb_checkpoints 79 | 80 | # IPython 81 | profile_default/ 82 | ipython_config.py 83 | 84 | # pyenv 85 | .python-version 86 | 87 | # pipenv 88 | # According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control. 89 | # However, in case of collaboration, if having platform-specific dependencies or dependencies 90 | # having no cross-platform support, pipenv may install dependencies that don't work, or not 91 | # install all needed dependencies. 92 | #Pipfile.lock 93 | 94 | # PEP 582; used by e.g. github.com/David-OConnor/pyflow 95 | __pypackages__/ 96 | 97 | # Celery stuff 98 | celerybeat-schedule 99 | celerybeat.pid 100 | 101 | # SageMath parsed files 102 | *.sage.py 103 | 104 | # Environments 105 | .env 106 | .venv 107 | env/ 108 | venv/ 109 | ENV/ 110 | env.bak/ 111 | venv.bak/ 112 | 113 | # Spyder project settings 114 | .spyderproject 115 | .spyproject 116 | 117 | # Rope project settings 118 | .ropeproject 119 | 120 | # mkdocs documentation 121 | /site 122 | 123 | # mypy 124 | .mypy_cache/ 125 | .dmypy.json 126 | dmypy.json 127 | 128 | # Pyre type checker 129 | .pyre/ 130 | -------------------------------------------------------------------------------- /ml_app/data_processing/pre_processor.py: -------------------------------------------------------------------------------- 1 | import pandas as pd 2 | import numpy as np 3 | 4 | 5 | def form_dataset(request_data, ml_model_artifacts, 6 | na_values=['', '?']): 7 | 8 | # convert request records into a list of dicts 9 | request_data = [request_data] if type(request_data) == dict else request_data 10 | # for each record add in missing fields 11 | for record in request_data: 12 | # get list of inital data features 13 | feature_names = list(ml_model_artifacts['cat_init_features']) + list(ml_model_artifacts['num_init_features']) 14 | # get list of features missing in record 15 | features_not_present = list(set(feature_names) - set(record.keys())) 16 | # fill feature names with a missing value placeholder 17 | for feature in features_not_present: 18 | record[feature] = '?' 19 | 20 | # convert list of record dicts into a dataframe 21 | request_df = pd.DataFrame(request_data) 22 | # convert missing value tokens to NaNs 23 | for token in na_values: 24 | request_df = request_df.replace({token : np.NaN}) 25 | 26 | return request_df 27 | 28 | 29 | def impute_and_encode_features(request_df, ml_model_artifacts): 30 | 31 | # separate categorical and numeric features 32 | categorical_features_init = ml_model_artifacts['cat_init_features'] 33 | numeric_features_init = ml_model_artifacts['num_init_features'] 34 | request_df_cat = request_df[categorical_features_init] 35 | request_df_num = request_df[numeric_features_init] 36 | 37 | # impute categorical features 38 | categorical_imputer = ml_model_artifacts['cat_imputer'] 39 | request_df_cat = pd.DataFrame(categorical_imputer.transform(request_df_cat), 40 | columns=categorical_features_init) 41 | 42 | # one-hot encode categorical features (dummy variables) 43 | categorical_ohe = ml_model_artifacts['dummy_encoder'] 44 | request_df_cat_ohe = categorical_ohe.transform(request_df_cat).toarray() 45 | 46 | categorical_features_ohe = ml_model_artifacts['cat_ohe_features'] 47 | request_df_cat_ohe = pd.DataFrame(request_df_cat_ohe, 48 | columns=categorical_features_ohe) 49 | 50 | # impute numeric features 51 | numeric_imputer = ml_model_artifacts['num_imputer'] 52 | request_df_num = pd.DataFrame(numeric_imputer.transform(request_df_num), 53 | columns=numeric_features_init) 54 | 55 | # combine numeric and categorical features 56 | request_df = pd.concat([request_df_num, request_df_cat_ohe], axis=1) 57 | # align column names for feature set 58 | column_names = ml_model_artifacts['column_names_order'] 59 | request_df = request_df[column_names] 60 | 61 | return request_df -------------------------------------------------------------------------------- /notebooks/Test ML API.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "markdown", 5 | "metadata": {}, 6 | "source": [ 7 | "# Load Dependencies" 8 | ] 9 | }, 10 | { 11 | "cell_type": "code", 12 | "execution_count": 20, 13 | "metadata": {}, 14 | "outputs": [], 15 | "source": [ 16 | "import requests\n", 17 | "import json" 18 | ] 19 | }, 20 | { 21 | "cell_type": "markdown", 22 | "metadata": {}, 23 | "source": [ 24 | "# API Liveness Test" 25 | ] 26 | }, 27 | { 28 | "cell_type": "code", 29 | "execution_count": 8, 30 | "metadata": {}, 31 | "outputs": [], 32 | "source": [ 33 | "API_LIVENESS_URL = 'http://ec2-44-192-74-26.compute-1.amazonaws.com:8900/income_classifier/api/v1/liveness'" 34 | ] 35 | }, 36 | { 37 | "cell_type": "code", 38 | "execution_count": 9, 39 | "metadata": {}, 40 | "outputs": [], 41 | "source": [ 42 | "response = requests.get(API_LIVENESS_URL)" 43 | ] 44 | }, 45 | { 46 | "cell_type": "code", 47 | "execution_count": 10, 48 | "metadata": {}, 49 | "outputs": [ 50 | { 51 | "data": { 52 | "text/plain": [ 53 | "'API Live!'" 54 | ] 55 | }, 56 | "execution_count": 10, 57 | "metadata": {}, 58 | "output_type": "execute_result" 59 | } 60 | ], 61 | "source": [ 62 | "response.text" 63 | ] 64 | }, 65 | { 66 | "cell_type": "markdown", 67 | "metadata": {}, 68 | "source": [ 69 | "# API ML Inference Test" 70 | ] 71 | }, 72 | { 73 | "cell_type": "code", 74 | "execution_count": 100, 75 | "metadata": {}, 76 | "outputs": [], 77 | "source": [ 78 | "import pandas as pd\n", 79 | "\n", 80 | "df = pd.read_csv('../datasets/census.csv')\n", 81 | "final_columns = set(df.columns) - set(['fnlwgt'])\n", 82 | "final_columns = list(final_columns)\n", 83 | "df = df[final_columns]\n", 84 | "df = df.drop(columns=['income'])\n", 85 | "\n", 86 | "data = df.iloc[20001:20010].to_dict(orient='records')" 87 | ] 88 | }, 89 | { 90 | "cell_type": "code", 91 | "execution_count": 101, 92 | "metadata": {}, 93 | "outputs": [], 94 | "source": [ 95 | "API_INFERENCE_URL = 'http://ec2-44-192-74-26.compute-1.amazonaws.com:8900/income_classifier/api/v1/predict'\n", 96 | "HEADERS = {'content-type': 'application/json'}" 97 | ] 98 | }, 99 | { 100 | "cell_type": "code", 101 | "execution_count": 102, 102 | "metadata": {}, 103 | "outputs": [ 104 | { 105 | "data": { 106 | "text/plain": [ 107 | "{'capital.gain': 0,\n", 108 | " 'sex': 'Male',\n", 109 | " 'capital.loss': 0,\n", 110 | " 'age': 37,\n", 111 | " 'relationship': 'Husband',\n", 112 | " 'hours.per.week': 40,\n", 113 | " 'education.num': 13,\n", 114 | " 'native.country': 'United-States',\n", 115 | " 'occupation': 'Exec-managerial',\n", 116 | " 'race': 'White',\n", 117 | " 'education': 'Bachelors',\n", 118 | " 'marital.status': 'Married-civ-spouse',\n", 119 | " 'workclass': 'Private'}" 120 | ] 121 | }, 122 | "execution_count": 102, 123 | "metadata": {}, 124 | "output_type": "execute_result" 125 | } 126 | ], 127 | "source": [ 128 | "data[0]" 129 | ] 130 | }, 131 | { 132 | "cell_type": "code", 133 | "execution_count": 103, 134 | "metadata": {}, 135 | "outputs": [ 136 | { 137 | "data": { 138 | "text/plain": [ 139 | "'{\"data\": {\"capital.gain\": 0, \"sex\": \"Male\", \"capital.loss\": 0, \"age\": 37, \"relationship\": \"Husband\", \"hours.per.week\": 40, \"education.num\": 13, \"native.country\": \"United-States\", \"occupation\": \"Exec-managerial\", \"race\": \"White\", \"education\": \"Bachelors\", \"marital.status\": \"Married-civ-spouse\", \"workclass\": \"Private\"}}'" 140 | ] 141 | }, 142 | "execution_count": 103, 143 | "metadata": {}, 144 | "output_type": "execute_result" 145 | } 146 | ], 147 | "source": [ 148 | "request_data = data[0]\n", 149 | "request = json.dumps({'data': request_data})\n", 150 | "request" 151 | ] 152 | }, 153 | { 154 | "cell_type": "code", 155 | "execution_count": 104, 156 | "metadata": {}, 157 | "outputs": [ 158 | { 159 | "data": { 160 | "text/plain": [ 161 | "{'predicted_classes': ['>50K']}" 162 | ] 163 | }, 164 | "execution_count": 104, 165 | "metadata": {}, 166 | "output_type": "execute_result" 167 | } 168 | ], 169 | "source": [ 170 | "json_response = requests.post(API_INFERENCE_URL, data=request, headers=HEADERS)\n", 171 | "json_response.json()" 172 | ] 173 | }, 174 | { 175 | "cell_type": "code", 176 | "execution_count": 115, 177 | "metadata": {}, 178 | "outputs": [ 179 | { 180 | "data": { 181 | "text/plain": [ 182 | "'{\"data\": [{\"capital.gain\": 0, \"sex\": \"Male\", \"capital.loss\": 0, \"age\": 58, \"relationship\": \"Husband\", \"hours.per.week\": 20, \"education.num\": 9, \"native.country\": \"United-States\", \"occupation\": \"Craft-repair\", \"race\": \"White\", \"education\": \"HS-grad\", \"marital.status\": \"Married-civ-spouse\", \"workclass\": \"Self-emp-not-inc\"}, {\"capital.gain\": 0, \"sex\": \"Female\", \"capital.loss\": 0, \"age\": 26, \"relationship\": \"Not-in-family\", \"hours.per.week\": 40, \"education.num\": 13, \"native.country\": \"United-States\", \"occupation\": \"Adm-clerical\", \"race\": \"Black\", \"education\": \"Bachelors\", \"marital.status\": \"Never-married\", \"workclass\": \"Private\"}, {\"capital.gain\": 0, \"sex\": \"Male\", \"capital.loss\": 0, \"age\": 55, \"relationship\": \"Husband\", \"hours.per.week\": 50, \"education.num\": 10, \"native.country\": \"Canada\", \"occupation\": \"Tech-support\", \"race\": \"White\", \"education\": \"Some-college\", \"marital.status\": \"Married-civ-spouse\", \"workclass\": \"Private\"}, {\"capital.gain\": 0, \"sex\": \"Male\", \"capital.loss\": 0, \"age\": 46, \"relationship\": \"Husband\", \"hours.per.week\": 48, \"education.num\": 10, \"native.country\": \"United-States\", \"occupation\": \"Exec-managerial\", \"race\": \"White\", \"education\": \"Some-college\", \"marital.status\": \"Married-civ-spouse\", \"workclass\": \"Federal-gov\"}, {\"capital.gain\": 0, \"sex\": \"Male\", \"capital.loss\": 0, \"age\": 44, \"relationship\": \"Husband\", \"hours.per.week\": 60, \"education.num\": 7, \"native.country\": \"United-States\", \"occupation\": \"Other-service\", \"race\": \"Black\", \"education\": \"11th\", \"marital.status\": \"Married-civ-spouse\", \"workclass\": \"Private\"}, {\"capital.gain\": 0, \"sex\": \"Male\", \"capital.loss\": 0, \"age\": 28, \"relationship\": \"Not-in-family\", \"hours.per.week\": 55, \"education.num\": 13, \"native.country\": \"United-States\", \"occupation\": \"Exec-managerial\", \"race\": \"White\", \"education\": \"Bachelors\", \"marital.status\": \"Never-married\", \"workclass\": \"State-gov\"}, {\"capital.gain\": 0, \"sex\": \"Male\", \"capital.loss\": 0, \"age\": 42, \"relationship\": \"Husband\", \"hours.per.week\": 50, \"education.num\": 6, \"native.country\": \"United-States\", \"occupation\": \"Transport-moving\", \"race\": \"White\", \"education\": \"10th\", \"marital.status\": \"Married-civ-spouse\", \"workclass\": \"Private\"}, {\"capital.gain\": 0, \"sex\": \"Male\", \"capital.loss\": 0, \"age\": 41, \"relationship\": \"Not-in-family\", \"hours.per.week\": 40, \"education.num\": 9, \"native.country\": \"United-States\", \"occupation\": \"Craft-repair\", \"race\": \"White\", \"education\": \"HS-grad\", \"marital.status\": \"Divorced\", \"workclass\": \"Private\"}]}'" 183 | ] 184 | }, 185 | "execution_count": 115, 186 | "metadata": {}, 187 | "output_type": "execute_result" 188 | } 189 | ], 190 | "source": [ 191 | "request_data = data[1:]\n", 192 | "request = json.dumps({'data': request_data})\n", 193 | "request" 194 | ] 195 | }, 196 | { 197 | "cell_type": "code", 198 | "execution_count": 116, 199 | "metadata": {}, 200 | "outputs": [ 201 | { 202 | "data": { 203 | "text/plain": [ 204 | "{'predicted_classes': ['<=50K',\n", 205 | " '<=50K',\n", 206 | " '>50K',\n", 207 | " '>50K',\n", 208 | " '<=50K',\n", 209 | " '<=50K',\n", 210 | " '<=50K',\n", 211 | " '<=50K']}" 212 | ] 213 | }, 214 | "execution_count": 116, 215 | "metadata": {}, 216 | "output_type": "execute_result" 217 | } 218 | ], 219 | "source": [ 220 | "json_response = requests.post(API_INFERENCE_URL, data=request, headers=HEADERS)\n", 221 | "json_response.json()" 222 | ] 223 | }, 224 | { 225 | "cell_type": "code", 226 | "execution_count": null, 227 | "metadata": {}, 228 | "outputs": [], 229 | "source": [] 230 | } 231 | ], 232 | "metadata": { 233 | "kernelspec": { 234 | "display_name": "Python 3", 235 | "language": "python", 236 | "name": "python3" 237 | }, 238 | "language_info": { 239 | "codemirror_mode": { 240 | "name": "ipython", 241 | "version": 3 242 | }, 243 | "file_extension": ".py", 244 | "mimetype": "text/x-python", 245 | "name": "python", 246 | "nbconvert_exporter": "python", 247 | "pygments_lexer": "ipython3", 248 | "version": "3.8.5" 249 | } 250 | }, 251 | "nbformat": 4, 252 | "nbformat_minor": 4 253 | } 254 | -------------------------------------------------------------------------------- /notebooks/ML Inference Pipeline.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "markdown", 5 | "metadata": {}, 6 | "source": [ 7 | "# Load Basic Dependencies" 8 | ] 9 | }, 10 | { 11 | "cell_type": "code", 12 | "execution_count": 95, 13 | "metadata": {}, 14 | "outputs": [ 15 | { 16 | "name": "stdout", 17 | "output_type": "stream", 18 | "text": [ 19 | "Pandas version 1.1.3\n", 20 | "Numpy version 1.19.2\n", 21 | "Scikit Learn version 0.23.2\n", 22 | "XGBoost version 1.4.1\n" 23 | ] 24 | } 25 | ], 26 | "source": [ 27 | "import pandas as pd\n", 28 | "import numpy as np\n", 29 | "import sklearn\n", 30 | "import xgboost as xgb\n", 31 | "\n", 32 | "print('Pandas version', pd.__version__)\n", 33 | "print('Numpy version', np.__version__)\n", 34 | "print('Scikit Learn version', sklearn.__version__)\n", 35 | "print('XGBoost version', xgb.__version__)" 36 | ] 37 | }, 38 | { 39 | "cell_type": "markdown", 40 | "metadata": {}, 41 | "source": [ 42 | "# Load Sample Data" 43 | ] 44 | }, 45 | { 46 | "cell_type": "code", 47 | "execution_count": 96, 48 | "metadata": {}, 49 | "outputs": [], 50 | "source": [ 51 | "df = pd.read_csv('../datasets/census.csv')\n", 52 | "final_columns = set(df.columns) - set(['fnlwgt'])\n", 53 | "final_columns = list(final_columns)\n", 54 | "df = df[final_columns]\n", 55 | "df = df.drop(columns=['income'])" 56 | ] 57 | }, 58 | { 59 | "cell_type": "markdown", 60 | "metadata": {}, 61 | "source": [ 62 | "# Create Sample Request Datasets\n", 63 | "\n", 64 | "Here we emulate how data would look when we build an API to serve model requests\n", 65 | "\n", 66 | "Typically requests and responses are generated in JSON, hence we will work with python dictionaries as inputs and outputs" 67 | ] 68 | }, 69 | { 70 | "cell_type": "code", 71 | "execution_count": 97, 72 | "metadata": {}, 73 | "outputs": [ 74 | { 75 | "data": { 76 | "text/plain": [ 77 | "{'education': 'Assoc-voc',\n", 78 | " 'workclass': 'Private',\n", 79 | " 'native.country': 'United-States',\n", 80 | " 'sex': 'Male',\n", 81 | " 'education.num': 11,\n", 82 | " 'race': 'White',\n", 83 | " 'occupation': 'Craft-repair',\n", 84 | " 'capital.gain': 0,\n", 85 | " 'capital.loss': 2603,\n", 86 | " 'marital.status': 'Married-civ-spouse',\n", 87 | " 'relationship': 'Husband',\n", 88 | " 'hours.per.week': 40,\n", 89 | " 'age': 21}" 90 | ] 91 | }, 92 | "execution_count": 97, 93 | "metadata": {}, 94 | "output_type": "execute_result" 95 | } 96 | ], 97 | "source": [ 98 | "request_data1 = df.iloc[25].to_dict()\n", 99 | "request_data1" 100 | ] 101 | }, 102 | { 103 | "cell_type": "markdown", 104 | "metadata": {}, 105 | "source": [ 106 | "Creating another sample request dataset with multiple records and introducing more missing data and fields to simulate real-world scenarios" 107 | ] 108 | }, 109 | { 110 | "cell_type": "code", 111 | "execution_count": 98, 112 | "metadata": {}, 113 | "outputs": [ 114 | { 115 | "data": { 116 | "text/plain": [ 117 | "[{'education': 'HS-grad',\n", 118 | " 'workclass': '?',\n", 119 | " 'occupation': '?',\n", 120 | " 'capital.gain': 0,\n", 121 | " 'capital.loss': 4356,\n", 122 | " 'hours.per.week': 40},\n", 123 | " {'education': 'HS-grad',\n", 124 | " 'workclass': 'Private',\n", 125 | " 'native.country': 'United-States',\n", 126 | " 'sex': 'Female',\n", 127 | " 'education.num': 9,\n", 128 | " 'race': 'White',\n", 129 | " 'occupation': 'Exec-managerial',\n", 130 | " 'capital.gain': 0,\n", 131 | " 'capital.loss': 4356,\n", 132 | " 'marital.status': 'Widowed',\n", 133 | " 'relationship': 'Not-in-family',\n", 134 | " 'hours.per.week': 18,\n", 135 | " 'age': 82},\n", 136 | " {'education': 'Some-college',\n", 137 | " 'workclass': '',\n", 138 | " 'native.country': 'United-States',\n", 139 | " 'sex': 'Female',\n", 140 | " 'education.num': 10,\n", 141 | " 'race': '?',\n", 142 | " 'occupation': '?',\n", 143 | " 'capital.gain': 0,\n", 144 | " 'capital.loss': 4356,\n", 145 | " 'marital.status': 'Widowed',\n", 146 | " 'relationship': 'Unmarried',\n", 147 | " 'hours.per.week': '?',\n", 148 | " 'age': 66}]" 149 | ] 150 | }, 151 | "execution_count": 98, 152 | "metadata": {}, 153 | "output_type": "execute_result" 154 | } 155 | ], 156 | "source": [ 157 | "request_data2 = df.iloc[0:3].to_dict(orient='records')\n", 158 | "\n", 159 | "request_data2[2]['workclass'] = ''\n", 160 | "request_data2[2]['race'] = '?'\n", 161 | "request_data2[2]['hours.per.week'] = '?'\n", 162 | "\n", 163 | "del request_data2[0]['native.country']\n", 164 | "del request_data2[0]['sex']\n", 165 | "del request_data2[0]['age']\n", 166 | "del request_data2[0]['race']\n", 167 | "del request_data2[0]['relationship']\n", 168 | "del request_data2[0]['marital.status']\n", 169 | "del request_data2[0]['education.num']\n", 170 | "\n", 171 | "request_data2" 172 | ] 173 | }, 174 | { 175 | "cell_type": "markdown", 176 | "metadata": {}, 177 | "source": [ 178 | "# Step 1: Create function to load model artifacts" 179 | ] 180 | }, 181 | { 182 | "cell_type": "code", 183 | "execution_count": 99, 184 | "metadata": {}, 185 | "outputs": [], 186 | "source": [ 187 | "import dill\n", 188 | "\n", 189 | "\n", 190 | "def load_model_artifacts(path):\n", 191 | " with open(path, \"rb\") as dill_infile:\n", 192 | " model_artifacts = dill.load(dill_infile)\n", 193 | " \n", 194 | " return model_artifacts" 195 | ] 196 | }, 197 | { 198 | "cell_type": "code", 199 | "execution_count": 100, 200 | "metadata": {}, 201 | "outputs": [ 202 | { 203 | "data": { 204 | "text/plain": [ 205 | "dict_keys(['dummy_encoder', 'cat_init_features', 'num_init_features', 'cat_ohe_features', 'cat_imputer', 'num_imputer', 'xgb_model', 'column_names_order'])" 206 | ] 207 | }, 208 | "execution_count": 100, 209 | "metadata": {}, 210 | "output_type": "execute_result" 211 | } 212 | ], 213 | "source": [ 214 | "ML_ARTIFACTS_PATH = \"../ml_app/saved_models/census_xgb_artifacts.pkl\"\n", 215 | "\n", 216 | "ml_artifacts = load_model_artifacts(path=ML_ARTIFACTS_PATH)\n", 217 | "ml_artifacts.keys()" 218 | ] 219 | }, 220 | { 221 | "cell_type": "markdown", 222 | "metadata": {}, 223 | "source": [ 224 | "# Step 2: Create function to form a dataset from request data" 225 | ] 226 | }, 227 | { 228 | "cell_type": "code", 229 | "execution_count": 101, 230 | "metadata": {}, 231 | "outputs": [], 232 | "source": [ 233 | "def form_dataset(request_data, ml_model_artifacts,\n", 234 | " na_values=['', '?']):\n", 235 | " \n", 236 | " # convert request records into a list of dicts\n", 237 | " request_data = [request_data] if type(request_data) == dict else request_data\n", 238 | " # for each record add in missing fields\n", 239 | " for record in request_data:\n", 240 | " # get list of inital data features\n", 241 | " feature_names = list(ml_model_artifacts['cat_init_features']) + list(ml_model_artifacts['num_init_features'])\n", 242 | " # get list of features missing in record\n", 243 | " features_not_present = list(set(feature_names) - set(record.keys()))\n", 244 | " # fill feature names with a missing value placeholder\n", 245 | " for feature in features_not_present:\n", 246 | " record[feature] = '?'\n", 247 | " \n", 248 | " # convert list of record dicts into a dataframe \n", 249 | " request_df = pd.DataFrame(request_data)\n", 250 | " # convert missing value tokens to NaNs\n", 251 | " for token in na_values:\n", 252 | " request_df = request_df.replace({token : np.NaN})\n", 253 | "\n", 254 | " return request_df" 255 | ] 256 | }, 257 | { 258 | "cell_type": "code", 259 | "execution_count": 102, 260 | "metadata": {}, 261 | "outputs": [ 262 | { 263 | "data": { 264 | "text/html": [ 265 | "
\n", 266 | "\n", 279 | "\n", 280 | " \n", 281 | " \n", 282 | " \n", 283 | " \n", 284 | " \n", 285 | " \n", 286 | " \n", 287 | " \n", 288 | " \n", 289 | " \n", 290 | " \n", 291 | " \n", 292 | " \n", 293 | " \n", 294 | " \n", 295 | " \n", 296 | " \n", 297 | " \n", 298 | " \n", 299 | " \n", 300 | " \n", 301 | " \n", 302 | " \n", 303 | " \n", 304 | " \n", 305 | " \n", 306 | " \n", 307 | " \n", 308 | " \n", 309 | " \n", 310 | " \n", 311 | " \n", 312 | " \n", 313 | " \n", 314 | " \n", 315 | " \n", 316 | " \n", 317 | " \n", 318 | " \n", 319 | " \n", 320 | " \n", 321 | " \n", 322 | " \n", 323 | " \n", 324 | " \n", 325 | " \n", 326 | " \n", 327 | " \n", 328 | " \n", 329 | " \n", 330 | " \n", 331 | " \n", 332 | " \n", 333 | " \n", 334 | " \n", 335 | " \n", 336 | " \n", 337 | " \n", 338 | " \n", 339 | " \n", 340 | " \n", 341 | " \n", 342 | " \n", 343 | " \n", 344 | " \n", 345 | " \n", 346 | " \n", 347 | " \n", 348 | "
educationworkclassoccupationcapital.gaincapital.losshours.per.weeknative.countrysexeducation.numracemarital.statusrelationshipage
0HS-gradNaNNaN0435640.0NaNNaNNaNNaNNaNNaNNaN
1HS-gradPrivateExec-managerial0435618.0United-StatesFemale9.0WhiteWidowedNot-in-family82.0
2Some-collegeNaNNaN04356NaNUnited-StatesFemale10.0NaNWidowedUnmarried66.0
\n", 349 | "
" 350 | ], 351 | "text/plain": [ 352 | " education workclass occupation capital.gain capital.loss \\\n", 353 | "0 HS-grad NaN NaN 0 4356 \n", 354 | "1 HS-grad Private Exec-managerial 0 4356 \n", 355 | "2 Some-college NaN NaN 0 4356 \n", 356 | "\n", 357 | " hours.per.week native.country sex education.num race marital.status \\\n", 358 | "0 40.0 NaN NaN NaN NaN NaN \n", 359 | "1 18.0 United-States Female 9.0 White Widowed \n", 360 | "2 NaN United-States Female 10.0 NaN Widowed \n", 361 | "\n", 362 | " relationship age \n", 363 | "0 NaN NaN \n", 364 | "1 Not-in-family 82.0 \n", 365 | "2 Unmarried 66.0 " 366 | ] 367 | }, 368 | "execution_count": 102, 369 | "metadata": {}, 370 | "output_type": "execute_result" 371 | } 372 | ], 373 | "source": [ 374 | "request_df = form_dataset(request_data=request_data2,\n", 375 | " ml_model_artifacts=ml_artifacts)\n", 376 | "request_df" 377 | ] 378 | }, 379 | { 380 | "cell_type": "markdown", 381 | "metadata": {}, 382 | "source": [ 383 | "# Step 3: Impute and Encode Features" 384 | ] 385 | }, 386 | { 387 | "cell_type": "code", 388 | "execution_count": 103, 389 | "metadata": {}, 390 | "outputs": [], 391 | "source": [ 392 | "def impute_and_encode_features(request_df, ml_model_artifacts):\n", 393 | " \n", 394 | " # separate categorical and numeric features\n", 395 | " categorical_features_init = ml_model_artifacts['cat_init_features']\n", 396 | " numeric_features_init = ml_model_artifacts['num_init_features']\n", 397 | " request_df_cat = request_df[categorical_features_init]\n", 398 | " request_df_num = request_df[numeric_features_init]\n", 399 | " \n", 400 | " # impute categorical features\n", 401 | " categorical_imputer = ml_model_artifacts['cat_imputer']\n", 402 | " request_df_cat = pd.DataFrame(categorical_imputer.transform(request_df_cat), \n", 403 | " columns=categorical_features_init)\n", 404 | " \n", 405 | " # one-hot encode categorical features (dummy variables)\n", 406 | " categorical_ohe = ml_model_artifacts['dummy_encoder']\n", 407 | " request_df_cat_ohe = categorical_ohe.transform(request_df_cat).toarray()\n", 408 | " \n", 409 | " categorical_features_ohe = ml_model_artifacts['cat_ohe_features']\n", 410 | " request_df_cat_ohe = pd.DataFrame(request_df_cat_ohe, \n", 411 | " columns=categorical_features_ohe)\n", 412 | " \n", 413 | " # impute numeric features\n", 414 | " numeric_imputer = ml_model_artifacts['num_imputer']\n", 415 | " request_df_num = pd.DataFrame(numeric_imputer.transform(request_df_num), \n", 416 | " columns=numeric_features_init)\n", 417 | " \n", 418 | " # combine numeric and categorical features\n", 419 | " request_df = pd.concat([request_df_num, request_df_cat_ohe], axis=1)\n", 420 | " # align column names for feature set\n", 421 | " column_names = ml_model_artifacts['column_names_order']\n", 422 | " request_df = request_df[column_names]\n", 423 | " \n", 424 | " return request_df" 425 | ] 426 | }, 427 | { 428 | "cell_type": "code", 429 | "execution_count": 104, 430 | "metadata": {}, 431 | "outputs": [ 432 | { 433 | "data": { 434 | "text/html": [ 435 | "
\n", 436 | "\n", 449 | "\n", 450 | " \n", 451 | " \n", 452 | " \n", 453 | " \n", 454 | " \n", 455 | " \n", 456 | " \n", 457 | " \n", 458 | " \n", 459 | " \n", 460 | " \n", 461 | " \n", 462 | " \n", 463 | " \n", 464 | " \n", 465 | " \n", 466 | " \n", 467 | " \n", 468 | " \n", 469 | " \n", 470 | " \n", 471 | " \n", 472 | " \n", 473 | " \n", 474 | " \n", 475 | " \n", 476 | " \n", 477 | " \n", 478 | " \n", 479 | " \n", 480 | " \n", 481 | " \n", 482 | " \n", 483 | " \n", 484 | " \n", 485 | " \n", 486 | " \n", 487 | " \n", 488 | " \n", 489 | " \n", 490 | " \n", 491 | " \n", 492 | " \n", 493 | " \n", 494 | " \n", 495 | " \n", 496 | " \n", 497 | " \n", 498 | " \n", 499 | " \n", 500 | " \n", 501 | " \n", 502 | " \n", 503 | " \n", 504 | " \n", 505 | " \n", 506 | " \n", 507 | " \n", 508 | " \n", 509 | " \n", 510 | " \n", 511 | " \n", 512 | " \n", 513 | " \n", 514 | " \n", 515 | " \n", 516 | " \n", 517 | " \n", 518 | " \n", 519 | " \n", 520 | " \n", 521 | " \n", 522 | " \n", 523 | " \n", 524 | " \n", 525 | " \n", 526 | " \n", 527 | " \n", 528 | " \n", 529 | " \n", 530 | " \n", 531 | " \n", 532 | " \n", 533 | " \n", 534 | " \n", 535 | " \n", 536 | " \n", 537 | " \n", 538 | " \n", 539 | " \n", 540 | " \n", 541 | " \n", 542 | " \n", 543 | " \n", 544 | " \n", 545 | " \n", 546 | " \n", 547 | " \n", 548 | " \n", 549 | " \n", 550 | "
agehours.per.weekeducation.numcapital.losscapital.gainrelationship_Husbandrelationship_Not-in-familyrelationship_Other-relativerelationship_Own-childrelationship_Unmarried...education_9theducation_Assoc-acdmeducation_Assoc-voceducation_Bachelorseducation_Doctorateeducation_HS-gradeducation_Masterseducation_Preschooleducation_Prof-schooleducation_Some-college
057.840.07.84356.00.00.00.00.00.00.0...0.00.00.00.00.01.00.00.00.00.0
182.018.09.04356.00.00.01.00.00.00.0...0.00.00.00.00.01.00.00.00.00.0
266.040.010.04356.00.00.00.00.00.01.0...0.00.00.00.00.00.00.00.00.01.0
\n", 551 | "

3 rows × 106 columns

\n", 552 | "
" 553 | ], 554 | "text/plain": [ 555 | " age hours.per.week education.num capital.loss capital.gain \\\n", 556 | "0 57.8 40.0 7.8 4356.0 0.0 \n", 557 | "1 82.0 18.0 9.0 4356.0 0.0 \n", 558 | "2 66.0 40.0 10.0 4356.0 0.0 \n", 559 | "\n", 560 | " relationship_Husband relationship_Not-in-family \\\n", 561 | "0 0.0 0.0 \n", 562 | "1 0.0 1.0 \n", 563 | "2 0.0 0.0 \n", 564 | "\n", 565 | " relationship_Other-relative relationship_Own-child \\\n", 566 | "0 0.0 0.0 \n", 567 | "1 0.0 0.0 \n", 568 | "2 0.0 0.0 \n", 569 | "\n", 570 | " relationship_Unmarried ... education_9th education_Assoc-acdm \\\n", 571 | "0 0.0 ... 0.0 0.0 \n", 572 | "1 0.0 ... 0.0 0.0 \n", 573 | "2 1.0 ... 0.0 0.0 \n", 574 | "\n", 575 | " education_Assoc-voc education_Bachelors education_Doctorate \\\n", 576 | "0 0.0 0.0 0.0 \n", 577 | "1 0.0 0.0 0.0 \n", 578 | "2 0.0 0.0 0.0 \n", 579 | "\n", 580 | " education_HS-grad education_Masters education_Preschool \\\n", 581 | "0 1.0 0.0 0.0 \n", 582 | "1 1.0 0.0 0.0 \n", 583 | "2 0.0 0.0 0.0 \n", 584 | "\n", 585 | " education_Prof-school education_Some-college \n", 586 | "0 0.0 0.0 \n", 587 | "1 0.0 0.0 \n", 588 | "2 0.0 1.0 \n", 589 | "\n", 590 | "[3 rows x 106 columns]" 591 | ] 592 | }, 593 | "execution_count": 104, 594 | "metadata": {}, 595 | "output_type": "execute_result" 596 | } 597 | ], 598 | "source": [ 599 | "request_df = impute_and_encode_features(request_df=request_df, \n", 600 | " ml_model_artifacts=ml_artifacts)\n", 601 | "request_df" 602 | ] 603 | }, 604 | { 605 | "cell_type": "markdown", 606 | "metadata": {}, 607 | "source": [ 608 | "# Step 4: Load and make ML model predictions" 609 | ] 610 | }, 611 | { 612 | "cell_type": "code", 613 | "execution_count": 105, 614 | "metadata": {}, 615 | "outputs": [], 616 | "source": [ 617 | "def make_model_predictions(request_df, ml_model_artifacts):\n", 618 | " \n", 619 | " # load saved ML model\n", 620 | " ml_model = ml_model_artifacts['xgb_model']\n", 621 | " \n", 622 | " # make model predictions\n", 623 | " predictions = ml_model.predict(request_df)\n", 624 | " \n", 625 | " # return predictions\n", 626 | " return {\n", 627 | " 'predicted_classes' : list(predictions)\n", 628 | " }" 629 | ] 630 | }, 631 | { 632 | "cell_type": "code", 633 | "execution_count": 106, 634 | "metadata": {}, 635 | "outputs": [ 636 | { 637 | "name": "stderr", 638 | "output_type": "stream", 639 | "text": [ 640 | "/opt/conda/lib/python3.8/site-packages/xgboost/data.py:112: UserWarning: Use subset (sliced data) of np.ndarray is not recommended because it will generate extra copies and increase memory consumption\n", 641 | " warnings.warn(\n" 642 | ] 643 | }, 644 | { 645 | "data": { 646 | "text/plain": [ 647 | "{'predicted_classes': ['<=50K', '<=50K', '<=50K']}" 648 | ] 649 | }, 650 | "execution_count": 106, 651 | "metadata": {}, 652 | "output_type": "execute_result" 653 | } 654 | ], 655 | "source": [ 656 | "make_model_predictions(request_df=request_df, \n", 657 | " ml_model_artifacts=ml_artifacts)" 658 | ] 659 | }, 660 | { 661 | "cell_type": "markdown", 662 | "metadata": {}, 663 | "source": [ 664 | "# Step 5: Build ML inference pipeline" 665 | ] 666 | }, 667 | { 668 | "cell_type": "code", 669 | "execution_count": 107, 670 | "metadata": {}, 671 | "outputs": [], 672 | "source": [ 673 | "ML_ARTIFACTS_PATH = \"../ml_app/saved_models/census_xgb_artifacts.pkl\"\n", 674 | "\n", 675 | "\n", 676 | "def ml_inference_pipeline(request_data):\n", 677 | " \n", 678 | " # 1. Load model artifacts\n", 679 | " ml_artifacts = load_model_artifacts(path=ML_ARTIFACTS_PATH)\n", 680 | " \n", 681 | " # 2. Create request dataset\n", 682 | " request_df = form_dataset(request_data=request_data,\n", 683 | " ml_model_artifacts=ml_artifacts)\n", 684 | " \n", 685 | " # 3. Impute and Encode Features\n", 686 | " request_df = impute_and_encode_features(request_df=request_df, \n", 687 | " ml_model_artifacts=ml_artifacts)\n", 688 | " \n", 689 | " # 4. Load and make ML model predictions\n", 690 | " pred_response = make_model_predictions(request_df=request_df, \n", 691 | " ml_model_artifacts=ml_artifacts)\n", 692 | " \n", 693 | " # return response\n", 694 | " return pred_response\n" 695 | ] 696 | }, 697 | { 698 | "cell_type": "markdown", 699 | "metadata": {}, 700 | "source": [ 701 | "# Test inference pipeline" 702 | ] 703 | }, 704 | { 705 | "cell_type": "code", 706 | "execution_count": 108, 707 | "metadata": {}, 708 | "outputs": [ 709 | { 710 | "data": { 711 | "text/plain": [ 712 | "{'predicted_classes': ['<=50K']}" 713 | ] 714 | }, 715 | "execution_count": 108, 716 | "metadata": {}, 717 | "output_type": "execute_result" 718 | } 719 | ], 720 | "source": [ 721 | "ml_inference_pipeline(request_data=request_data1)" 722 | ] 723 | }, 724 | { 725 | "cell_type": "code", 726 | "execution_count": 109, 727 | "metadata": {}, 728 | "outputs": [ 729 | { 730 | "name": "stderr", 731 | "output_type": "stream", 732 | "text": [ 733 | "/opt/conda/lib/python3.8/site-packages/xgboost/data.py:112: UserWarning: Use subset (sliced data) of np.ndarray is not recommended because it will generate extra copies and increase memory consumption\n", 734 | " warnings.warn(\n" 735 | ] 736 | }, 737 | { 738 | "data": { 739 | "text/plain": [ 740 | "{'predicted_classes': ['<=50K', '<=50K', '<=50K']}" 741 | ] 742 | }, 743 | "execution_count": 109, 744 | "metadata": {}, 745 | "output_type": "execute_result" 746 | } 747 | ], 748 | "source": [ 749 | "ml_inference_pipeline(request_data=request_data2)" 750 | ] 751 | }, 752 | { 753 | "cell_type": "code", 754 | "execution_count": 110, 755 | "metadata": {}, 756 | "outputs": [], 757 | "source": [ 758 | "df_raw = pd.read_csv('../datasets/census.csv')" 759 | ] 760 | }, 761 | { 762 | "cell_type": "code", 763 | "execution_count": 111, 764 | "metadata": {}, 765 | "outputs": [], 766 | "source": [ 767 | "request_data3 = df.iloc[20000:20010].to_dict(orient='records')" 768 | ] 769 | }, 770 | { 771 | "cell_type": "code", 772 | "execution_count": 112, 773 | "metadata": {}, 774 | "outputs": [ 775 | { 776 | "name": "stderr", 777 | "output_type": "stream", 778 | "text": [ 779 | "/opt/conda/lib/python3.8/site-packages/xgboost/data.py:112: UserWarning: Use subset (sliced data) of np.ndarray is not recommended because it will generate extra copies and increase memory consumption\n", 780 | " warnings.warn(\n" 781 | ] 782 | }, 783 | { 784 | "data": { 785 | "text/plain": [ 786 | "{'predicted_classes': ['<=50K',\n", 787 | " '>50K',\n", 788 | " '<=50K',\n", 789 | " '<=50K',\n", 790 | " '>50K',\n", 791 | " '>50K',\n", 792 | " '<=50K',\n", 793 | " '<=50K',\n", 794 | " '<=50K',\n", 795 | " '<=50K']}" 796 | ] 797 | }, 798 | "execution_count": 112, 799 | "metadata": {}, 800 | "output_type": "execute_result" 801 | } 802 | ], 803 | "source": [ 804 | "ml_inference_pipeline(request_data=request_data3)" 805 | ] 806 | }, 807 | { 808 | "cell_type": "code", 809 | "execution_count": 113, 810 | "metadata": {}, 811 | "outputs": [ 812 | { 813 | "data": { 814 | "text/plain": [ 815 | "['<=50K',\n", 816 | " '>50K',\n", 817 | " '<=50K',\n", 818 | " '<=50K',\n", 819 | " '>50K',\n", 820 | " '>50K',\n", 821 | " '<=50K',\n", 822 | " '<=50K',\n", 823 | " '<=50K',\n", 824 | " '<=50K']" 825 | ] 826 | }, 827 | "execution_count": 113, 828 | "metadata": {}, 829 | "output_type": "execute_result" 830 | } 831 | ], 832 | "source": [ 833 | "df_raw.iloc[20000:20010]['income'].tolist()" 834 | ] 835 | } 836 | ], 837 | "metadata": { 838 | "kernelspec": { 839 | "display_name": "Python 3", 840 | "language": "python", 841 | "name": "python3" 842 | }, 843 | "language_info": { 844 | "codemirror_mode": { 845 | "name": "ipython", 846 | "version": 3 847 | }, 848 | "file_extension": ".py", 849 | "mimetype": "text/x-python", 850 | "name": "python", 851 | "nbconvert_exporter": "python", 852 | "pygments_lexer": "ipython3", 853 | "version": "3.8.5" 854 | } 855 | }, 856 | "nbformat": 4, 857 | "nbformat_minor": 4 858 | } 859 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | GNU GENERAL PUBLIC LICENSE 2 | Version 3, 29 June 2007 3 | 4 | Copyright (C) 2007 Free Software Foundation, Inc. 5 | Everyone is permitted to copy and distribute verbatim copies 6 | of this license document, but changing it is not allowed. 7 | 8 | Preamble 9 | 10 | The GNU General Public License is a free, copyleft license for 11 | software and other kinds of works. 12 | 13 | The licenses for most software and other practical works are designed 14 | to take away your freedom to share and change the works. By contrast, 15 | the GNU General Public License is intended to guarantee your freedom to 16 | share and change all versions of a program--to make sure it remains free 17 | software for all its users. We, the Free Software Foundation, use the 18 | GNU General Public License for most of our software; it applies also to 19 | any other work released this way by its authors. You can apply it to 20 | your programs, too. 21 | 22 | When we speak of free software, we are referring to freedom, not 23 | price. Our General Public Licenses are designed to make sure that you 24 | have the freedom to distribute copies of free software (and charge for 25 | them if you wish), that you receive source code or can get it if you 26 | want it, that you can change the software or use pieces of it in new 27 | free programs, and that you know you can do these things. 28 | 29 | To protect your rights, we need to prevent others from denying you 30 | these rights or asking you to surrender the rights. Therefore, you have 31 | certain responsibilities if you distribute copies of the software, or if 32 | you modify it: responsibilities to respect the freedom of others. 33 | 34 | For example, if you distribute copies of such a program, whether 35 | gratis or for a fee, you must pass on to the recipients the same 36 | freedoms that you received. You must make sure that they, too, receive 37 | or can get the source code. And you must show them these terms so they 38 | know their rights. 39 | 40 | Developers that use the GNU GPL protect your rights with two steps: 41 | (1) assert copyright on the software, and (2) offer you this License 42 | giving you legal permission to copy, distribute and/or modify it. 43 | 44 | For the developers' and authors' protection, the GPL clearly explains 45 | that there is no warranty for this free software. For both users' and 46 | authors' sake, the GPL requires that modified versions be marked as 47 | changed, so that their problems will not be attributed erroneously to 48 | authors of previous versions. 49 | 50 | Some devices are designed to deny users access to install or run 51 | modified versions of the software inside them, although the manufacturer 52 | can do so. This is fundamentally incompatible with the aim of 53 | protecting users' freedom to change the software. The systematic 54 | pattern of such abuse occurs in the area of products for individuals to 55 | use, which is precisely where it is most unacceptable. Therefore, we 56 | have designed this version of the GPL to prohibit the practice for those 57 | products. If such problems arise substantially in other domains, we 58 | stand ready to extend this provision to those domains in future versions 59 | of the GPL, as needed to protect the freedom of users. 60 | 61 | Finally, every program is threatened constantly by software patents. 62 | States should not allow patents to restrict development and use of 63 | software on general-purpose computers, but in those that do, we wish to 64 | avoid the special danger that patents applied to a free program could 65 | make it effectively proprietary. To prevent this, the GPL assures that 66 | patents cannot be used to render the program non-free. 67 | 68 | The precise terms and conditions for copying, distribution and 69 | modification follow. 70 | 71 | TERMS AND CONDITIONS 72 | 73 | 0. Definitions. 74 | 75 | "This License" refers to version 3 of the GNU General Public License. 76 | 77 | "Copyright" also means copyright-like laws that apply to other kinds of 78 | works, such as semiconductor masks. 79 | 80 | "The Program" refers to any copyrightable work licensed under this 81 | License. Each licensee is addressed as "you". "Licensees" and 82 | "recipients" may be individuals or organizations. 83 | 84 | To "modify" a work means to copy from or adapt all or part of the work 85 | in a fashion requiring copyright permission, other than the making of an 86 | exact copy. The resulting work is called a "modified version" of the 87 | earlier work or a work "based on" the earlier work. 88 | 89 | A "covered work" means either the unmodified Program or a work based 90 | on the Program. 91 | 92 | To "propagate" a work means to do anything with it that, without 93 | permission, would make you directly or secondarily liable for 94 | infringement under applicable copyright law, except executing it on a 95 | computer or modifying a private copy. Propagation includes copying, 96 | distribution (with or without modification), making available to the 97 | public, and in some countries other activities as well. 98 | 99 | To "convey" a work means any kind of propagation that enables other 100 | parties to make or receive copies. Mere interaction with a user through 101 | a computer network, with no transfer of a copy, is not conveying. 102 | 103 | An interactive user interface displays "Appropriate Legal Notices" 104 | to the extent that it includes a convenient and prominently visible 105 | feature that (1) displays an appropriate copyright notice, and (2) 106 | tells the user that there is no warranty for the work (except to the 107 | extent that warranties are provided), that licensees may convey the 108 | work under this License, and how to view a copy of this License. If 109 | the interface presents a list of user commands or options, such as a 110 | menu, a prominent item in the list meets this criterion. 111 | 112 | 1. Source Code. 113 | 114 | The "source code" for a work means the preferred form of the work 115 | for making modifications to it. "Object code" means any non-source 116 | form of a work. 117 | 118 | A "Standard Interface" means an interface that either is an official 119 | standard defined by a recognized standards body, or, in the case of 120 | interfaces specified for a particular programming language, one that 121 | is widely used among developers working in that language. 122 | 123 | The "System Libraries" of an executable work include anything, other 124 | than the work as a whole, that (a) is included in the normal form of 125 | packaging a Major Component, but which is not part of that Major 126 | Component, and (b) serves only to enable use of the work with that 127 | Major Component, or to implement a Standard Interface for which an 128 | implementation is available to the public in source code form. A 129 | "Major Component", in this context, means a major essential component 130 | (kernel, window system, and so on) of the specific operating system 131 | (if any) on which the executable work runs, or a compiler used to 132 | produce the work, or an object code interpreter used to run it. 133 | 134 | The "Corresponding Source" for a work in object code form means all 135 | the source code needed to generate, install, and (for an executable 136 | work) run the object code and to modify the work, including scripts to 137 | control those activities. However, it does not include the work's 138 | System Libraries, or general-purpose tools or generally available free 139 | programs which are used unmodified in performing those activities but 140 | which are not part of the work. For example, Corresponding Source 141 | includes interface definition files associated with source files for 142 | the work, and the source code for shared libraries and dynamically 143 | linked subprograms that the work is specifically designed to require, 144 | such as by intimate data communication or control flow between those 145 | subprograms and other parts of the work. 146 | 147 | The Corresponding Source need not include anything that users 148 | can regenerate automatically from other parts of the Corresponding 149 | Source. 150 | 151 | The Corresponding Source for a work in source code form is that 152 | same work. 153 | 154 | 2. Basic Permissions. 155 | 156 | All rights granted under this License are granted for the term of 157 | copyright on the Program, and are irrevocable provided the stated 158 | conditions are met. This License explicitly affirms your unlimited 159 | permission to run the unmodified Program. The output from running a 160 | covered work is covered by this License only if the output, given its 161 | content, constitutes a covered work. This License acknowledges your 162 | rights of fair use or other equivalent, as provided by copyright law. 163 | 164 | You may make, run and propagate covered works that you do not 165 | convey, without conditions so long as your license otherwise remains 166 | in force. You may convey covered works to others for the sole purpose 167 | of having them make modifications exclusively for you, or provide you 168 | with facilities for running those works, provided that you comply with 169 | the terms of this License in conveying all material for which you do 170 | not control copyright. Those thus making or running the covered works 171 | for you must do so exclusively on your behalf, under your direction 172 | and control, on terms that prohibit them from making any copies of 173 | your copyrighted material outside their relationship with you. 174 | 175 | Conveying under any other circumstances is permitted solely under 176 | the conditions stated below. Sublicensing is not allowed; section 10 177 | makes it unnecessary. 178 | 179 | 3. Protecting Users' Legal Rights From Anti-Circumvention Law. 180 | 181 | No covered work shall be deemed part of an effective technological 182 | measure under any applicable law fulfilling obligations under article 183 | 11 of the WIPO copyright treaty adopted on 20 December 1996, or 184 | similar laws prohibiting or restricting circumvention of such 185 | measures. 186 | 187 | When you convey a covered work, you waive any legal power to forbid 188 | circumvention of technological measures to the extent such circumvention 189 | is effected by exercising rights under this License with respect to 190 | the covered work, and you disclaim any intention to limit operation or 191 | modification of the work as a means of enforcing, against the work's 192 | users, your or third parties' legal rights to forbid circumvention of 193 | technological measures. 194 | 195 | 4. Conveying Verbatim Copies. 196 | 197 | You may convey verbatim copies of the Program's source code as you 198 | receive it, in any medium, provided that you conspicuously and 199 | appropriately publish on each copy an appropriate copyright notice; 200 | keep intact all notices stating that this License and any 201 | non-permissive terms added in accord with section 7 apply to the code; 202 | keep intact all notices of the absence of any warranty; and give all 203 | recipients a copy of this License along with the Program. 204 | 205 | You may charge any price or no price for each copy that you convey, 206 | and you may offer support or warranty protection for a fee. 207 | 208 | 5. Conveying Modified Source Versions. 209 | 210 | You may convey a work based on the Program, or the modifications to 211 | produce it from the Program, in the form of source code under the 212 | terms of section 4, provided that you also meet all of these conditions: 213 | 214 | a) The work must carry prominent notices stating that you modified 215 | it, and giving a relevant date. 216 | 217 | b) The work must carry prominent notices stating that it is 218 | released under this License and any conditions added under section 219 | 7. This requirement modifies the requirement in section 4 to 220 | "keep intact all notices". 221 | 222 | c) You must license the entire work, as a whole, under this 223 | License to anyone who comes into possession of a copy. This 224 | License will therefore apply, along with any applicable section 7 225 | additional terms, to the whole of the work, and all its parts, 226 | regardless of how they are packaged. This License gives no 227 | permission to license the work in any other way, but it does not 228 | invalidate such permission if you have separately received it. 229 | 230 | d) If the work has interactive user interfaces, each must display 231 | Appropriate Legal Notices; however, if the Program has interactive 232 | interfaces that do not display Appropriate Legal Notices, your 233 | work need not make them do so. 234 | 235 | A compilation of a covered work with other separate and independent 236 | works, which are not by their nature extensions of the covered work, 237 | and which are not combined with it such as to form a larger program, 238 | in or on a volume of a storage or distribution medium, is called an 239 | "aggregate" if the compilation and its resulting copyright are not 240 | used to limit the access or legal rights of the compilation's users 241 | beyond what the individual works permit. Inclusion of a covered work 242 | in an aggregate does not cause this License to apply to the other 243 | parts of the aggregate. 244 | 245 | 6. Conveying Non-Source Forms. 246 | 247 | You may convey a covered work in object code form under the terms 248 | of sections 4 and 5, provided that you also convey the 249 | machine-readable Corresponding Source under the terms of this License, 250 | in one of these ways: 251 | 252 | a) Convey the object code in, or embodied in, a physical product 253 | (including a physical distribution medium), accompanied by the 254 | Corresponding Source fixed on a durable physical medium 255 | customarily used for software interchange. 256 | 257 | b) Convey the object code in, or embodied in, a physical product 258 | (including a physical distribution medium), accompanied by a 259 | written offer, valid for at least three years and valid for as 260 | long as you offer spare parts or customer support for that product 261 | model, to give anyone who possesses the object code either (1) a 262 | copy of the Corresponding Source for all the software in the 263 | product that is covered by this License, on a durable physical 264 | medium customarily used for software interchange, for a price no 265 | more than your reasonable cost of physically performing this 266 | conveying of source, or (2) access to copy the 267 | Corresponding Source from a network server at no charge. 268 | 269 | c) Convey individual copies of the object code with a copy of the 270 | written offer to provide the Corresponding Source. This 271 | alternative is allowed only occasionally and noncommercially, and 272 | only if you received the object code with such an offer, in accord 273 | with subsection 6b. 274 | 275 | d) Convey the object code by offering access from a designated 276 | place (gratis or for a charge), and offer equivalent access to the 277 | Corresponding Source in the same way through the same place at no 278 | further charge. You need not require recipients to copy the 279 | Corresponding Source along with the object code. If the place to 280 | copy the object code is a network server, the Corresponding Source 281 | may be on a different server (operated by you or a third party) 282 | that supports equivalent copying facilities, provided you maintain 283 | clear directions next to the object code saying where to find the 284 | Corresponding Source. Regardless of what server hosts the 285 | Corresponding Source, you remain obligated to ensure that it is 286 | available for as long as needed to satisfy these requirements. 287 | 288 | e) Convey the object code using peer-to-peer transmission, provided 289 | you inform other peers where the object code and Corresponding 290 | Source of the work are being offered to the general public at no 291 | charge under subsection 6d. 292 | 293 | A separable portion of the object code, whose source code is excluded 294 | from the Corresponding Source as a System Library, need not be 295 | included in conveying the object code work. 296 | 297 | A "User Product" is either (1) a "consumer product", which means any 298 | tangible personal property which is normally used for personal, family, 299 | or household purposes, or (2) anything designed or sold for incorporation 300 | into a dwelling. In determining whether a product is a consumer product, 301 | doubtful cases shall be resolved in favor of coverage. For a particular 302 | product received by a particular user, "normally used" refers to a 303 | typical or common use of that class of product, regardless of the status 304 | of the particular user or of the way in which the particular user 305 | actually uses, or expects or is expected to use, the product. A product 306 | is a consumer product regardless of whether the product has substantial 307 | commercial, industrial or non-consumer uses, unless such uses represent 308 | the only significant mode of use of the product. 309 | 310 | "Installation Information" for a User Product means any methods, 311 | procedures, authorization keys, or other information required to install 312 | and execute modified versions of a covered work in that User Product from 313 | a modified version of its Corresponding Source. The information must 314 | suffice to ensure that the continued functioning of the modified object 315 | code is in no case prevented or interfered with solely because 316 | modification has been made. 317 | 318 | If you convey an object code work under this section in, or with, or 319 | specifically for use in, a User Product, and the conveying occurs as 320 | part of a transaction in which the right of possession and use of the 321 | User Product is transferred to the recipient in perpetuity or for a 322 | fixed term (regardless of how the transaction is characterized), the 323 | Corresponding Source conveyed under this section must be accompanied 324 | by the Installation Information. But this requirement does not apply 325 | if neither you nor any third party retains the ability to install 326 | modified object code on the User Product (for example, the work has 327 | been installed in ROM). 328 | 329 | The requirement to provide Installation Information does not include a 330 | requirement to continue to provide support service, warranty, or updates 331 | for a work that has been modified or installed by the recipient, or for 332 | the User Product in which it has been modified or installed. Access to a 333 | network may be denied when the modification itself materially and 334 | adversely affects the operation of the network or violates the rules and 335 | protocols for communication across the network. 336 | 337 | Corresponding Source conveyed, and Installation Information provided, 338 | in accord with this section must be in a format that is publicly 339 | documented (and with an implementation available to the public in 340 | source code form), and must require no special password or key for 341 | unpacking, reading or copying. 342 | 343 | 7. Additional Terms. 344 | 345 | "Additional permissions" are terms that supplement the terms of this 346 | License by making exceptions from one or more of its conditions. 347 | Additional permissions that are applicable to the entire Program shall 348 | be treated as though they were included in this License, to the extent 349 | that they are valid under applicable law. If additional permissions 350 | apply only to part of the Program, that part may be used separately 351 | under those permissions, but the entire Program remains governed by 352 | this License without regard to the additional permissions. 353 | 354 | When you convey a copy of a covered work, you may at your option 355 | remove any additional permissions from that copy, or from any part of 356 | it. (Additional permissions may be written to require their own 357 | removal in certain cases when you modify the work.) You may place 358 | additional permissions on material, added by you to a covered work, 359 | for which you have or can give appropriate copyright permission. 360 | 361 | Notwithstanding any other provision of this License, for material you 362 | add to a covered work, you may (if authorized by the copyright holders of 363 | that material) supplement the terms of this License with terms: 364 | 365 | a) Disclaiming warranty or limiting liability differently from the 366 | terms of sections 15 and 16 of this License; or 367 | 368 | b) Requiring preservation of specified reasonable legal notices or 369 | author attributions in that material or in the Appropriate Legal 370 | Notices displayed by works containing it; or 371 | 372 | c) Prohibiting misrepresentation of the origin of that material, or 373 | requiring that modified versions of such material be marked in 374 | reasonable ways as different from the original version; or 375 | 376 | d) Limiting the use for publicity purposes of names of licensors or 377 | authors of the material; or 378 | 379 | e) Declining to grant rights under trademark law for use of some 380 | trade names, trademarks, or service marks; or 381 | 382 | f) Requiring indemnification of licensors and authors of that 383 | material by anyone who conveys the material (or modified versions of 384 | it) with contractual assumptions of liability to the recipient, for 385 | any liability that these contractual assumptions directly impose on 386 | those licensors and authors. 387 | 388 | All other non-permissive additional terms are considered "further 389 | restrictions" within the meaning of section 10. If the Program as you 390 | received it, or any part of it, contains a notice stating that it is 391 | governed by this License along with a term that is a further 392 | restriction, you may remove that term. If a license document contains 393 | a further restriction but permits relicensing or conveying under this 394 | License, you may add to a covered work material governed by the terms 395 | of that license document, provided that the further restriction does 396 | not survive such relicensing or conveying. 397 | 398 | If you add terms to a covered work in accord with this section, you 399 | must place, in the relevant source files, a statement of the 400 | additional terms that apply to those files, or a notice indicating 401 | where to find the applicable terms. 402 | 403 | Additional terms, permissive or non-permissive, may be stated in the 404 | form of a separately written license, or stated as exceptions; 405 | the above requirements apply either way. 406 | 407 | 8. Termination. 408 | 409 | You may not propagate or modify a covered work except as expressly 410 | provided under this License. Any attempt otherwise to propagate or 411 | modify it is void, and will automatically terminate your rights under 412 | this License (including any patent licenses granted under the third 413 | paragraph of section 11). 414 | 415 | However, if you cease all violation of this License, then your 416 | license from a particular copyright holder is reinstated (a) 417 | provisionally, unless and until the copyright holder explicitly and 418 | finally terminates your license, and (b) permanently, if the copyright 419 | holder fails to notify you of the violation by some reasonable means 420 | prior to 60 days after the cessation. 421 | 422 | Moreover, your license from a particular copyright holder is 423 | reinstated permanently if the copyright holder notifies you of the 424 | violation by some reasonable means, this is the first time you have 425 | received notice of violation of this License (for any work) from that 426 | copyright holder, and you cure the violation prior to 30 days after 427 | your receipt of the notice. 428 | 429 | Termination of your rights under this section does not terminate the 430 | licenses of parties who have received copies or rights from you under 431 | this License. If your rights have been terminated and not permanently 432 | reinstated, you do not qualify to receive new licenses for the same 433 | material under section 10. 434 | 435 | 9. Acceptance Not Required for Having Copies. 436 | 437 | You are not required to accept this License in order to receive or 438 | run a copy of the Program. Ancillary propagation of a covered work 439 | occurring solely as a consequence of using peer-to-peer transmission 440 | to receive a copy likewise does not require acceptance. However, 441 | nothing other than this License grants you permission to propagate or 442 | modify any covered work. These actions infringe copyright if you do 443 | not accept this License. Therefore, by modifying or propagating a 444 | covered work, you indicate your acceptance of this License to do so. 445 | 446 | 10. Automatic Licensing of Downstream Recipients. 447 | 448 | Each time you convey a covered work, the recipient automatically 449 | receives a license from the original licensors, to run, modify and 450 | propagate that work, subject to this License. You are not responsible 451 | for enforcing compliance by third parties with this License. 452 | 453 | An "entity transaction" is a transaction transferring control of an 454 | organization, or substantially all assets of one, or subdividing an 455 | organization, or merging organizations. If propagation of a covered 456 | work results from an entity transaction, each party to that 457 | transaction who receives a copy of the work also receives whatever 458 | licenses to the work the party's predecessor in interest had or could 459 | give under the previous paragraph, plus a right to possession of the 460 | Corresponding Source of the work from the predecessor in interest, if 461 | the predecessor has it or can get it with reasonable efforts. 462 | 463 | You may not impose any further restrictions on the exercise of the 464 | rights granted or affirmed under this License. For example, you may 465 | not impose a license fee, royalty, or other charge for exercise of 466 | rights granted under this License, and you may not initiate litigation 467 | (including a cross-claim or counterclaim in a lawsuit) alleging that 468 | any patent claim is infringed by making, using, selling, offering for 469 | sale, or importing the Program or any portion of it. 470 | 471 | 11. Patents. 472 | 473 | A "contributor" is a copyright holder who authorizes use under this 474 | License of the Program or a work on which the Program is based. The 475 | work thus licensed is called the contributor's "contributor version". 476 | 477 | A contributor's "essential patent claims" are all patent claims 478 | owned or controlled by the contributor, whether already acquired or 479 | hereafter acquired, that would be infringed by some manner, permitted 480 | by this License, of making, using, or selling its contributor version, 481 | but do not include claims that would be infringed only as a 482 | consequence of further modification of the contributor version. For 483 | purposes of this definition, "control" includes the right to grant 484 | patent sublicenses in a manner consistent with the requirements of 485 | this License. 486 | 487 | Each contributor grants you a non-exclusive, worldwide, royalty-free 488 | patent license under the contributor's essential patent claims, to 489 | make, use, sell, offer for sale, import and otherwise run, modify and 490 | propagate the contents of its contributor version. 491 | 492 | In the following three paragraphs, a "patent license" is any express 493 | agreement or commitment, however denominated, not to enforce a patent 494 | (such as an express permission to practice a patent or covenant not to 495 | sue for patent infringement). To "grant" such a patent license to a 496 | party means to make such an agreement or commitment not to enforce a 497 | patent against the party. 498 | 499 | If you convey a covered work, knowingly relying on a patent license, 500 | and the Corresponding Source of the work is not available for anyone 501 | to copy, free of charge and under the terms of this License, through a 502 | publicly available network server or other readily accessible means, 503 | then you must either (1) cause the Corresponding Source to be so 504 | available, or (2) arrange to deprive yourself of the benefit of the 505 | patent license for this particular work, or (3) arrange, in a manner 506 | consistent with the requirements of this License, to extend the patent 507 | license to downstream recipients. "Knowingly relying" means you have 508 | actual knowledge that, but for the patent license, your conveying the 509 | covered work in a country, or your recipient's use of the covered work 510 | in a country, would infringe one or more identifiable patents in that 511 | country that you have reason to believe are valid. 512 | 513 | If, pursuant to or in connection with a single transaction or 514 | arrangement, you convey, or propagate by procuring conveyance of, a 515 | covered work, and grant a patent license to some of the parties 516 | receiving the covered work authorizing them to use, propagate, modify 517 | or convey a specific copy of the covered work, then the patent license 518 | you grant is automatically extended to all recipients of the covered 519 | work and works based on it. 520 | 521 | A patent license is "discriminatory" if it does not include within 522 | the scope of its coverage, prohibits the exercise of, or is 523 | conditioned on the non-exercise of one or more of the rights that are 524 | specifically granted under this License. You may not convey a covered 525 | work if you are a party to an arrangement with a third party that is 526 | in the business of distributing software, under which you make payment 527 | to the third party based on the extent of your activity of conveying 528 | the work, and under which the third party grants, to any of the 529 | parties who would receive the covered work from you, a discriminatory 530 | patent license (a) in connection with copies of the covered work 531 | conveyed by you (or copies made from those copies), or (b) primarily 532 | for and in connection with specific products or compilations that 533 | contain the covered work, unless you entered into that arrangement, 534 | or that patent license was granted, prior to 28 March 2007. 535 | 536 | Nothing in this License shall be construed as excluding or limiting 537 | any implied license or other defenses to infringement that may 538 | otherwise be available to you under applicable patent law. 539 | 540 | 12. No Surrender of Others' Freedom. 541 | 542 | If conditions are imposed on you (whether by court order, agreement or 543 | otherwise) that contradict the conditions of this License, they do not 544 | excuse you from the conditions of this License. If you cannot convey a 545 | covered work so as to satisfy simultaneously your obligations under this 546 | License and any other pertinent obligations, then as a consequence you may 547 | not convey it at all. For example, if you agree to terms that obligate you 548 | to collect a royalty for further conveying from those to whom you convey 549 | the Program, the only way you could satisfy both those terms and this 550 | License would be to refrain entirely from conveying the Program. 551 | 552 | 13. Use with the GNU Affero General Public License. 553 | 554 | Notwithstanding any other provision of this License, you have 555 | permission to link or combine any covered work with a work licensed 556 | under version 3 of the GNU Affero General Public License into a single 557 | combined work, and to convey the resulting work. The terms of this 558 | License will continue to apply to the part which is the covered work, 559 | but the special requirements of the GNU Affero General Public License, 560 | section 13, concerning interaction through a network will apply to the 561 | combination as such. 562 | 563 | 14. Revised Versions of this License. 564 | 565 | The Free Software Foundation may publish revised and/or new versions of 566 | the GNU General Public License from time to time. Such new versions will 567 | be similar in spirit to the present version, but may differ in detail to 568 | address new problems or concerns. 569 | 570 | Each version is given a distinguishing version number. If the 571 | Program specifies that a certain numbered version of the GNU General 572 | Public License "or any later version" applies to it, you have the 573 | option of following the terms and conditions either of that numbered 574 | version or of any later version published by the Free Software 575 | Foundation. If the Program does not specify a version number of the 576 | GNU General Public License, you may choose any version ever published 577 | by the Free Software Foundation. 578 | 579 | If the Program specifies that a proxy can decide which future 580 | versions of the GNU General Public License can be used, that proxy's 581 | public statement of acceptance of a version permanently authorizes you 582 | to choose that version for the Program. 583 | 584 | Later license versions may give you additional or different 585 | permissions. However, no additional obligations are imposed on any 586 | author or copyright holder as a result of your choosing to follow a 587 | later version. 588 | 589 | 15. Disclaimer of Warranty. 590 | 591 | THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY 592 | APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT 593 | HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY 594 | OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, 595 | THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 596 | PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM 597 | IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF 598 | ALL NECESSARY SERVICING, REPAIR OR CORRECTION. 599 | 600 | 16. Limitation of Liability. 601 | 602 | IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING 603 | WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS 604 | THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY 605 | GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE 606 | USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF 607 | DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD 608 | PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS), 609 | EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF 610 | SUCH DAMAGES. 611 | 612 | 17. Interpretation of Sections 15 and 16. 613 | 614 | If the disclaimer of warranty and limitation of liability provided 615 | above cannot be given local legal effect according to their terms, 616 | reviewing courts shall apply local law that most closely approximates 617 | an absolute waiver of all civil liability in connection with the 618 | Program, unless a warranty or assumption of liability accompanies a 619 | copy of the Program in return for a fee. 620 | 621 | END OF TERMS AND CONDITIONS 622 | 623 | How to Apply These Terms to Your New Programs 624 | 625 | If you develop a new program, and you want it to be of the greatest 626 | possible use to the public, the best way to achieve this is to make it 627 | free software which everyone can redistribute and change under these terms. 628 | 629 | To do so, attach the following notices to the program. It is safest 630 | to attach them to the start of each source file to most effectively 631 | state the exclusion of warranty; and each file should have at least 632 | the "copyright" line and a pointer to where the full notice is found. 633 | 634 | 635 | Copyright (C) 636 | 637 | This program is free software: you can redistribute it and/or modify 638 | it under the terms of the GNU General Public License as published by 639 | the Free Software Foundation, either version 3 of the License, or 640 | (at your option) any later version. 641 | 642 | This program is distributed in the hope that it will be useful, 643 | but WITHOUT ANY WARRANTY; without even the implied warranty of 644 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 645 | GNU General Public License for more details. 646 | 647 | You should have received a copy of the GNU General Public License 648 | along with this program. If not, see . 649 | 650 | Also add information on how to contact you by electronic and paper mail. 651 | 652 | If the program does terminal interaction, make it output a short 653 | notice like this when it starts in an interactive mode: 654 | 655 | Copyright (C) 656 | This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'. 657 | This is free software, and you are welcome to redistribute it 658 | under certain conditions; type `show c' for details. 659 | 660 | The hypothetical commands `show w' and `show c' should show the appropriate 661 | parts of the General Public License. Of course, your program's commands 662 | might be different; for a GUI interface, you would use an "about box". 663 | 664 | You should also get your employer (if you work as a programmer) or school, 665 | if any, to sign a "copyright disclaimer" for the program, if necessary. 666 | For more information on this, and how to apply and follow the GNU GPL, see 667 | . 668 | 669 | The GNU General Public License does not permit incorporating your program 670 | into proprietary programs. If your program is a subroutine library, you 671 | may consider it more useful to permit linking proprietary applications with 672 | the library. If this is what you want to do, use the GNU Lesser General 673 | Public License instead of this License. But first, please read 674 | . 675 | --------------------------------------------------------------------------------