├── README.md └── notebooks └── bag-rf-var.ipynb /README.md: -------------------------------------------------------------------------------- 1 | # This repos is depreciated, check out the latest [Nailing Machine Learning Concepts](https://github.com/jayinai/nail-ml-concept) 2 | 3 | This repository covers how to prepare for machine learning interviews, mainly 4 | in the format of questions & answers. Asides from machine learning knowledge, 5 | other crucial aspects include: 6 | 7 | * [Explain your resume](#explain-your-resume) 8 | * [SQL](#sql) 9 | 10 | Go directly to [machine learning](#machine-learning) 11 | 12 | 13 | ## Explain your resume 14 | 15 | Your resume should specify interesting ML projects you got involved in the past, 16 | and **quantitatively** show your contribution. Consider the following comparison: 17 | 18 | > Trained a machine learning system 19 | 20 | vs. 21 | 22 | > Trained a deep vision system (SqueezeNet) that has 1/30 model size, 1/3 training 23 | > time, 1/5 inference time, and 2x faster convergence compared with traditional 24 | > ConvNet (e.g., ResNet) 25 | 26 | We all can tell which one is gonna catch interviewer's eyeballs and better show 27 | case your ability. 28 | 29 | In the interview, be sure to explain what you've done well. Spend some time going 30 | over your resume before the interview. 31 | 32 | 33 | ## SQL 34 | 35 | Although you don't have to be a SQL expert for most machine learning positions, 36 | the interviews might ask you some SQL related questions so it helps to refresh 37 | your memory beforehand. Some good SQL resources are: 38 | 39 | * [W3schools SQL](https://www.w3schools.com/sql/) 40 | * [SQLZOO](http://sqlzoo.net/) 41 | 42 | 43 | ## Machine learning 44 | 45 | First, it's always a good idea to review [Chapter 5](http://www.deeplearningbook.org/contents/ml.html) 46 | of the deep learning book, which covers machine learning basics. 47 | 48 | 49 | * [Linear regression](#linear-regression) 50 | * [Logistic regression](#logistic-regression) 51 | * [KNN](#knn) 52 | * [SVM](#svm) 53 | * [Naive Bayes] 54 | * [Decision tree](#decision-tree) 55 | * [Bagging](#bagging) 56 | * [Random forest](#random-forest) 57 | * [Boosting](#boosting) 58 | * [Stacking](#stacking) 59 | * [Clustering] 60 | * [MLP](#mlp) 61 | * [CNN](#cnn) 62 | * [RNN and LSTM](#rnn-and-lstm) 63 | * [word2vec](#word2vec) 64 | * [Generative vs discriminative](#generative-vs-discriminative) 65 | * [Paramteric vs Nonparametric](#paramteric-vs-nonparametric) 66 | 67 | 68 | 69 | ### Linear regression 70 | 71 | * how to learn the parameter: minimize the cost function 72 | * how to minimize cost function: gradient descent 73 | * regularization: 74 | - L1 (lasso): can shrink certain coef to zero, thus performing feature selection 75 | - L2 (ridge): shrink all coef with the same proportion; almost always outperforms L1 76 | - combined (Elastic Net): 77 | * assumes linear relationship between features and the label 78 | * can add polynomial and interaction features to add non-linearity 79 | 80 | ![lr](http://scikit-learn.org/stable/_images/sphx_glr_plot_cv_predict_thumb.png) 81 | 82 | [back to top](#machine-learning) 83 | 84 | 85 | ### Logistic regression 86 | 87 | * Generalized linear model (GLM) for classification problems 88 | * Apply the sigmoid function to the output of linear models, squeezing the target 89 | to range [0, 1] 90 | * Threshold to make prediction: if the output > .5, prediction 1; otherwise prediction 0 91 | * a special case of softmax function, which deals with multi-class problem 92 | 93 | [back to top](#machine-learning) 94 | 95 | ### KNN 96 | 97 | Given a data point, we compute the K nearest data points (neighbors) using certain 98 | distance metric (e.g., Euclidean metric). For classification, we take the majority label 99 | of neighbors; for regression, we take the mean of the label values. 100 | 101 | Note for KNN technically we don't need to train a model, we simply compute during 102 | inference time. This can be computationally expensive since each of the test example 103 | need to be compared with every training example to see how close they are. 104 | 105 | There are approximation methods can have faster inference time by 106 | partitioning the training data into regions. 107 | 108 | Note when K equals 1 or other small number the model is prone to overfitting (high variance), while 109 | when K equals number of data points or other large number the model is prone to underfitting (high bias) 110 | 111 | ![KNN](https://cambridgecoding.files.wordpress.com/2016/03/training_data_only_99_1.png?w=610) 112 | 113 | [back to top](#machine-learning) 114 | 115 | 116 | ### SVM 117 | 118 | * can perform linear, nonlinear, or outlier detection (unsupervised) 119 | * large margin classifier: not only have a decision boundary, but want the boundary 120 | to be as far from the closest training point as possible 121 | * the closest training examples are called the support vectors, since they are the points 122 | based on which the decision boundary is drawn 123 | * SVMs are sensitive to feature scaling 124 | 125 | ![svm](https://qph.ec.quoracdn.net/main-qimg-675fedee717331e478ecfcc40e2e4d38) 126 | 127 | 128 | [back to top](#machine-learning) 129 | 130 | 131 | ### Decision tree 132 | 133 | * Non-parametric, supervised learning algorithms 134 | * Given the training data, a decision tree algorithm divides the feature space into 135 | regions. For inference, we first see which 136 | region does the test data point fall in, and take the mean label values (regression) 137 | or the majority label value (classification). 138 | * **Construction**: top-down, chooses a variable to split the data such that the 139 | target variables within each region are as homogeneous as possible. Two common 140 | metrics: gini impurity or information gain, won't matter much in practice. 141 | * Advantage: simply to understand & interpret, mirrors human decision making 142 | * Disadvantage: 143 | - can overfit easily (and generalize poorly)if we don't limit the depth of the tree 144 | - can be non-robust: A small change in the training data can lead to a totally different tree 145 | - instability: sensitive to training set rotation due to its orthogonal decision boundaries 146 | 147 | ![decision tree](http://www.fizyka.umk.pl/~wduch/ref/kdd-tut/d-tree-iris.gif) 148 | 149 | [back to top](#machine-learning) 150 | 151 | 152 | ### Bagging 153 | 154 | To address overfitting, we can use an ensemble method called bagging (bootstrap aggregating), 155 | which reduces the variance of the meta learning algorithm. Bagging can be applied 156 | to decision tree or other algorithms. 157 | 158 | Here is a [great illustration](http://scikit-learn.org/stable/auto_examples/ensemble/plot_bias_variance.html#sphx-glr-auto-examples-ensemble-plot-bias-variance-py) of a single estimator vs. bagging 159 | 160 | ![bagging](http://scikit-learn.org/stable/_images/sphx_glr_plot_bias_variance_001.png) 161 | 162 | * Bagging is when samlping is performed *with* replacement. When sampling is performed *without* replacement, it's called pasting. 163 | * Bagging is popular due to its boost for performance, but also due to that individual learners can be trained in parallel and scale well 164 | * Ensemble methods work best when the learners are as independent from one another as possible 165 | * Voting: soft voting (predict probability and average over all individual learners) often works better than hard voting 166 | * out-of-bag instances (37%) can act validation set for bagging 167 | 168 | 169 | 170 | [back to top](#machine-learning) 171 | 172 | 173 | ### Random forest 174 | 175 | Random forest improves bagging further by adding some randomness. In random forest, 176 | only a subset of features are selected at random to construct a tree (while often not subsample instances). 177 | The benefit is that random forest **decorrelates** the trees. 178 | 179 | For example, suppose we have a dataset. There is one very predicative feature, and a couple 180 | of moderately predicative features. In bagging trees, most of the trees 181 | will use this very predicative feature in the top split, and therefore making most of the trees 182 | look similar, **and highly correlated**. Averaging many highly correlated results won't lead 183 | to a large reduction in variance compared with uncorrelated results. 184 | In random forest for each split we only consider a subset of the features and therefore 185 | reduce the variance even further by introducing more uncorrelated trees. 186 | 187 | I wrote a [notebook](notebooks/bag-rf-var.ipynb) to illustrate this point. 188 | 189 | In practice, tuning random forest entails having a large number of trees (the more the better, but 190 | always consider computation constraint). Also, `min_samples_leaf` (The minimum number of 191 | samples at the leaf node)to control the tree size and overfitting. Always CV the parameters. 192 | 193 | **Feature importance** 194 | 195 | In a decision tree, important features are likely to appear closer to the root of the tree. We can get 196 | a feature's importance for random forest by computing the averaging depth at which it appears across all 197 | trees in the forest. 198 | 199 | 200 | [back to top](#machine-learning) 201 | 202 | 203 | ### Boosting 204 | 205 | **How it works** 206 | 207 | Boosting builds on weak learners, and in an iterative fashion. In each iteration, 208 | a new learner is added, while all existing learners are kept unchanged. All learners 209 | are weighted based on their performance (e.g., accuracy), and after a weak learner 210 | is added, the data are re-weighted: examples that are misclassified gain more weights, 211 | while examples that are correctly classified lose weights. Thus, future weak learners 212 | focus more on examples that previous weak learners misclassified. 213 | 214 | 215 | **Difference from random forest (RF)** 216 | 217 | * RF grows trees **in parallel**, while Boosting is sequential 218 | * RF reduces variance, while Boosting reduces errors by reducing bias 219 | 220 | 221 | **XGBoost (Extreme Gradient Boosting)** 222 | 223 | 224 | > XGBoost uses a more regularized model formalization to control overfitting, which gives it better performance 225 | 226 | [back to top](#machine-learning) 227 | 228 | 229 | ### Stacking 230 | 231 | * Instead of using trivial functions (such as hard voting) to aggregate the predictions from individual learners, train a model to perform this aggregation 232 | * First split the training set into two subsets: the first subset is used to train the learners in the first layer 233 | * Next the first layer learners are used to make predictions (meta features) on the second subset, and those predictions are used to train another models (to obtain the weigts of different learners) in the second layer 234 | * We can train multiple models in the second layer, but this entails subsetting the original dataset into 3 parts 235 | 236 | ![stacking](http://www.kdnuggets.com/wp-content/uploads/backward-propagation-stacker-models.jpg) 237 | 238 | [back to top](#machine-learning) 239 | 240 | 241 | ### MLP 242 | 243 | A feedforward neural network where we have multiple layers. In each layer we 244 | can have multiple neurons, and each of the neuron in the next layer is a linear/nonlinear 245 | combination of the all the neurons in the previous layer. In order to train the network 246 | we back propagate the errors layer by layer. In theory MLP can approximate any functions. 247 | 248 | ![mlp](http://neuroph.sourceforge.net/tutorials/images/MLP.jpg) 249 | 250 | [back to top](#machine-learning) 251 | 252 | ### CNN 253 | 254 | The Conv layer is the building block of a Convolutional Network. The Conv layer consists 255 | of a set of learnable filters (such as 5 * 5 * 3, width * height * depth). During the forward 256 | pass, we slide (or more precisely, convolve) the filter across the input and compute the dot 257 | product. Learning again happens when the network back propagate the error layer by layer. 258 | 259 | Initial layers capture low-level features such as angle and edges, while later 260 | layers learn a combination of the low-level features and in the previous layers 261 | and can therefore represent higher level feature, such as shape and object parts. 262 | 263 | ![CNN](http://www.kdnuggets.com/wp-content/uploads/dnn-layers.jpg) 264 | 265 | [back to top](#machine-learning) 266 | 267 | ### RNN and LSTM 268 | 269 | RNN is another paradigm of neural network where we have difference layers of cells, 270 | and each cell only take as input the cell from the previous layer, but also the previous 271 | cell within the same layer. This gives RNN the power to model sequence. 272 | 273 | ![RNN](http://karpathy.github.io/assets/rnn/diags.jpeg) 274 | 275 | This seems great, but in practice RNN barely works due to exploding/vanishing gradient, which 276 | is cause by a series of multiplication of the same matrix. To solve this, we can use 277 | a variation of RNN, called long short-term memory (LSTM), which is capable of learning 278 | long-term dependencies. 279 | 280 | The math behind LSTM can be pretty complicated, but intuitively LSTM introduce 281 | - input gate 282 | - output gate 283 | - forget gate 284 | - memory cell (internal state) 285 | 286 | LSTM resembles human memory: it forgets old stuff (old internal state * forget gate) 287 | and learns from new input (input node * input gate) 288 | 289 | ![lstm](http://deeplearning.net/tutorial/_images/lstm_memorycell.png) 290 | 291 | [back to top](#machine-learning) 292 | 293 | 294 | ### word2vec 295 | 296 | * Shallow, two-layer neural networks that are trained to construct linguistic context of words 297 | * Takes as input a large corpus, and produce a vector space, typically of several hundred 298 | dimension, and each word in the corpus is assigned a vector in the space 299 | * The key idea is context: words that occur often in the same context should have same/opposite 300 | meanings. 301 | * Two flavors 302 | - continuous bag of words (CBOW): the model predicts the current word given a window of surrounding context words 303 | - skip gram: predicts the surrounding context words using the current word 304 | 305 | ![word2vec](https://deeplearning4j.org/img/countries_capitals.png) 306 | 307 | [back to top](#machine-learning) 308 | 309 | 310 | ### Generative vs discriminative 311 | 312 | * Discriminative algorithms model *p(y|x; w)*, that is, given the dataset and learned 313 | parameter, what is the probability of y belonging to a specific class. A discriminative algorithm 314 | doesn't care about how the data was generated, it simply categorizes a given example 315 | * Generative algorithms try to model *p(x|y)*, that is, the distribution of features given 316 | that it belongs to a certain class. A generative algorithm models how the data was 317 | generated. 318 | 319 | > Given a training set, an algorithm like logistic regression or 320 | > the perceptron algorithm (basically) tries to find a straight line—that is, a 321 | > decision boundary—that separates the elephants and dogs. Then, to classify 322 | > a new animal as either an elephant or a dog, it checks on which side of the 323 | > decision boundary it falls, and makes its prediction accordingly. 324 | 325 | > Here’s a different approach. First, looking at elephants, we can build a 326 | > model of what elephants look like. Then, looking at dogs, we can build a 327 | > separate model of what dogs look like. Finally, to classify a new animal, we 328 | > can match the new animal against the elephant model, and match it against 329 | > the dog model, to see whether the new animal looks more like the elephants 330 | > or more like the dogs we had seen in the training set. 331 | 332 | [back to top](#machine-learning) 333 | 334 | 335 | ### Paramteric vs Nonparametric 336 | 337 | * A learning model that summarizes data with a set of parameters of fixed size (independent of the number of training examples) is called a parametric model. 338 | * A model where the number of parameters is not determined prior to training. Nonparametric does not mean that they have NO parameters! On the contrary, nonparametric models (can) become more and more complex with an increasing amount of data. 339 | 340 | [back to top](#machine-learning) 341 | -------------------------------------------------------------------------------- /notebooks/bag-rf-var.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "code", 5 | "execution_count": 1, 6 | "metadata": { 7 | "collapsed": false 8 | }, 9 | "outputs": [], 10 | "source": [ 11 | "import numpy as np\n", 12 | "from sklearn.datasets import load_boston\n", 13 | "from sklearn.ensemble import BaggingRegressor, RandomForestRegressor\n", 14 | "from sklearn.model_selection import train_test_split\n", 15 | "import matplotlib.pyplot as plt\n", 16 | "%matplotlib inline" 17 | ] 18 | }, 19 | { 20 | "cell_type": "code", 21 | "execution_count": 2, 22 | "metadata": { 23 | "collapsed": true 24 | }, 25 | "outputs": [], 26 | "source": [ 27 | "boston = load_boston()\n", 28 | "X, y = boston.data, boston.target\n", 29 | "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=2017)" 30 | ] 31 | }, 32 | { 33 | "cell_type": "code", 34 | "execution_count": 3, 35 | "metadata": { 36 | "collapsed": false 37 | }, 38 | "outputs": [], 39 | "source": [ 40 | "n_tree = 50\n", 41 | "\n", 42 | "bag = BaggingRegressor(n_estimators=n_tree, n_jobs=-1)\n", 43 | "rf = RandomForestRegressor(n_estimators=n_tree, n_jobs=-1)\n", 44 | "\n", 45 | "bag.fit(X_train, y_train)\n", 46 | "rf.fit(X_train, y_train)\n", 47 | "\n", 48 | "bag_pred = np.array([bag.estimators_[tree].predict(X_test) for tree in range(n_tree)])\n", 49 | "rf_pred = np.array([rf.estimators_[tree].predict(X_test) for tree in range(n_tree)])\n", 50 | "\n", 51 | "bag_var = np.var(bag_pred, axis=0)\n", 52 | "rf_var = np.var(rf_pred, axis=0)" 53 | ] 54 | }, 55 | { 56 | "cell_type": "code", 57 | "execution_count": 4, 58 | "metadata": { 59 | "collapsed": false, 60 | "scrolled": false 61 | }, 62 | "outputs": [ 63 | { 64 | "data": { 65 | "image/png": "iVBORw0KGgoAAAANSUhEUgAAAq0AAAGyCAYAAADONP14AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n", 66 | "text/plain": [ 67 | "" 68 | ] 69 | }, 70 | "metadata": {}, 71 | "output_type": "display_data" 72 | } 73 | ], 74 | "source": [ 75 | "plt.figure(figsize=(8, 5))\n", 76 | "\n", 77 | "plt.plot(range(1, len(bag_var)+1), bag_var)\n", 78 | "plt.plot(range(1, len(rf_var)+1), rf_var)\n", 79 | "plt.legend(['bag_var', 'rf_var'])\n", 80 | "plt.show()" 81 | ] 82 | }, 83 | { 84 | "cell_type": "code", 85 | "execution_count": 5, 86 | "metadata": { 87 | "collapsed": false 88 | }, 89 | "outputs": [ 90 | { 91 | "name": "stdout", 92 | "output_type": "stream", 93 | "text": [ 94 | "bagging variance average: 15.121\n", 95 | "random forest variance average: 14.311\n" 96 | ] 97 | } 98 | ], 99 | "source": [ 100 | "print 'bagging variance average:', round(np.mean(bag_var), 3)\n", 101 | "print 'random forest variance average:', round(np.mean(rf_var), 3)" 102 | ] 103 | } 104 | ], 105 | "metadata": { 106 | "anaconda-cloud": {}, 107 | "kernelspec": { 108 | "display_name": "Python [default]", 109 | "language": "python", 110 | "name": "python2" 111 | }, 112 | "language_info": { 113 | "codemirror_mode": { 114 | "name": "ipython", 115 | "version": 2.0 116 | }, 117 | "file_extension": ".py", 118 | "mimetype": "text/x-python", 119 | "name": "python", 120 | "nbconvert_exporter": "python", 121 | "pygments_lexer": "ipython2", 122 | "version": "2.7.9" 123 | } 124 | }, 125 | "nbformat": 4, 126 | "nbformat_minor": 0 127 | } --------------------------------------------------------------------------------