{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "# Ensemble Methods Advanced concepts" ] }, { "cell_type": "code", "execution_count": 1, "metadata": { "collapsed": false, "scrolled": true }, "outputs": [ { "data": { "text/html": [ "
" ], "text/plain": [ "" ] }, "execution_count": 1, "metadata": {}, "output_type": "execute_result" } ], "source": [ "import addutils.toc ; addutils.toc.js(ipy_notebook=True)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "***Important*: install package `liac-arff` (use** `pip install liac-arff`**) in your environment
\n", "Do not use and uninstall the package `arff` if it's already installed in your environment**" ] }, { "cell_type": "code", "execution_count": 2, "metadata": { "collapsed": false }, "outputs": [ { "data": { "text/html": [ "\n", "\n" ], "text/plain": [ "" ] }, "execution_count": 2, "metadata": {}, "output_type": "execute_result" } ], "source": [ "import scipy.io\n", "import numpy as np\n", "import pandas as pd\n", "from time import time\n", "from sklearn import grid_search, cross_validation, metrics, ensemble, datasets\n", "import arff\n", "from addutils import css_notebook\n", "import sys\n", "import os\n", "import warnings\n", "css_notebook()" ] }, { "cell_type": "code", "execution_count": 3, "metadata": { "collapsed": false }, "outputs": [ { "data": { "text/html": [ " \n", "\n", "\n", " \n", "\n", "
\n", " \n", " BokehJS successfully loaded.\n", "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "import bokeh.plotting as bk\n", "bk.output_notebook()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## 1 Introduction" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "In this notebook we introduce some advanced techniques that can be used effectively with ensemble methods:\n", "\n", "- OOB Score (Estimates)\n", "- Feature Importance\n", "- Partial Dependece Plot" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## 2 Out of Bag Estimates" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "When using **ensemble methods** based upon **B**ootstrap **agg**regation or **Bagging**, i.e. using sampling with replacement, a part of the training set remains unused. For each classifier in the ensemble, a different part of the training set is left out. This left out portion can be used to **estimate the generalization error without having to rely on a separate Validation Set**. This estimate comes “for free” as no additional data is needed and can be used for model selection or for tuning hyperparameter.\n", "\n", "This is currently implemented in the following classes:\n", "\n", " * `RandomForestClassifier` / `RandomForestRegressor`\n", " * `ExtraTreesClassifier` / `ExtraTreesRegressor`\n", " * `GradientBoostingClassifier` / `GradientBoostingRegressor` \n", "\n", "**Remarks**\n", "\n", "In GradientBoosting OOB score is available only when `subsample` $< 1.0$. It can be used for a different purpose, that is finding the optimal number of boosting iterations, but it is a pessimistic estimator of the true test error. *Use it only for saving computataional time since, for example, cross-validation is more demanding*." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### 2.1 Phishing Websites Data Set\n", "For testing OOB Score properties of ensemble method we used a publicly available dataset. The dataset is the Phishing Websites Data Set and can be downloaded from [here](https://archive.ics.uci.edu/ml/datasets/Phishing+Websites). This dataset is anonymized by replacing the actual strings with describing values.\n", "\n", "The dataset is available in Weka format, so we have to use the *Weka arff file type reader for python* to read it." ] }, { "cell_type": "code", "execution_count": 4, "metadata": { "collapsed": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Keys: \n", "- data\n", "- relation\n", "- description\n", "- attributes\n" ] } ], "source": [ "#load the dataset\n", "with open('example_data/Training Dataset.arff', 'rb') as f:\n", " file_ = f.read().decode('utf-8')\n", "f.close()\n", "#print(file_)\n", "data = arff.load(file_)\n", "print(\"Keys: \")\n", "for k in data.keys():\n", " print(\"- \" + k)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "In \"attributes\" are stored the column names while actual data is stored in \"data\". The dataset is further divided into training and test set to compare OOB score with CV and held out set." ] }, { "cell_type": "code", "execution_count": 5, "metadata": { "collapsed": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Shape of the data: (11055, 31)\n", "Training set / Test set number of samples: 8844, 2211\n", "Number of features: 30\n" ] } ], "source": [ "#use the dataset to build a dataframe\n", "cols = [at[0] for at in data['attributes']]\n", "df = pd.DataFrame(data['data'], dtype=int, columns=cols)\n", "print(\"Shape of the data: {0}\".format(df.shape))\n", "target = df.iloc[:,-1]\n", "train = df.iloc[:,:-1]\n", "#split the dataframe in train / test\n", "X_train, X_test, y_train, y_test = cross_validation.train_test_split(train, \n", " target, \n", " test_size=0.2)\n", "print(\"Training set / Test set number of samples: {0}, {1}\".format(X_train.shape[0], \n", " X_test.shape[0]))\n", "print(\"Number of features: {0}\".format(X_train.shape[1]))" ] }, { "cell_type": "code", "execution_count": 6, "metadata": { "collapsed": false }, "outputs": [ { "data": { "text/html": [ "
\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
having_IP_AddressURL_LengthShortining_Service...Links_pointing_to_pageStatistical_reportResult
0-111...1-1-1
1111...11-1
2101...0-1-1
\n", "

3 rows × 31 columns

\n", "
" ], "text/plain": [ " having_IP_Address URL_Length Shortining_Service ... \\\n", "0 -1 1 1 ... \n", "1 1 1 1 ... \n", "2 1 0 1 ... \n", "\n", " Links_pointing_to_page Statistical_report Result \n", "0 1 -1 -1 \n", "1 1 1 -1 \n", "2 0 -1 -1 \n", "\n", "[3 rows x 31 columns]" ] }, "execution_count": 6, "metadata": {}, "output_type": "execute_result" } ], "source": [ "pd.options.display.max_columns=6\n", "df.head(3)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### 2.2 OOB estimate Random Forest\n", "Test OOB score against cross-validation and test set. In order to perform OOB score calculation the option oob_score must be set to True." ] }, { "cell_type": "code", "execution_count": 7, "metadata": { "collapsed": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "RF fit done in 9.666[s]\n", "CV score calculated in 63.303[s]\n" ] } ], "source": [ "t0 = time()\n", "rfc = ensemble.RandomForestClassifier(n_estimators=1000, \n", " oob_score=True, \n", " random_state=42,\n", " n_jobs=-1)\n", "rfc.fit(X_train, y_train)\n", "print('RF fit done in %0.3f[s]' %(time() - t0))\n", "t0 = time()\n", "cv_score = cross_validation.cross_val_score(rfc, train, target, cv=10, n_jobs=-1)\n", "print('CV score calculated in %0.3f[s]' %(time() - t0))" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "As can be seen in this example, using CV to measure the score takes time and it is similar to that obtained from OOB samples, that comes with the algorithm at no additional computational costs." ] }, { "cell_type": "code", "execution_count": 8, "metadata": { "collapsed": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Score comparison:\n", "F1 score (on the test set): 0.978910\n", "OOB score: 0.969923\n", "CV score (on the entire training set): 0.973038\n" ] } ], "source": [ "print('Score comparison:')\n", "print('F1 score (on the test set):'.ljust(40), '%f' % metrics.f1_score(rfc.predict(X_test),\\\n", " y_test))\n", "print('OOB score:'.ljust(40), '%f' % rfc.oob_score_)\n", "print('CV score (on the entire training set):'.ljust(40), '%f' % cv_score.mean())" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### 2.3 OOB estimate Gradient Boosting" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "**Please be patient: the following cell will require up to 10 minutes to be computed**" ] }, { "cell_type": "code", "execution_count": 9, "metadata": { "collapsed": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "GBRT fit done in 329.692[s]\n", "Accuracy: 0.9747\n", "CV score estimate done in 697.259[s]\n" ] } ], "source": [ "# Fit classifier with out-of-bag estimates\n", "params = {'n_estimators': 11000, \n", " 'max_depth': 3, \n", " 'subsample': 0.5,\n", " 'learning_rate': 0.1, \n", " 'min_samples_leaf': 1, \n", " 'random_state': 3}\n", "clf = ensemble.GradientBoostingClassifier(**params)\n", "\n", "t0 = time()\n", "clf.fit(X_train, y_train)\n", "print('GBRT fit done in %0.3f[s]' %(time() - t0))\n", "acc = clf.score(X_test, y_test)\n", "print(\"Accuracy: {:.4f}\".format(acc))\n", "\n", "n_estimators = params['n_estimators']\n", "x = np.arange(n_estimators) + 1\n", "\n", "def heldout_score(clf, X_test, y_test):\n", " \"\"\"compute deviance scores on ``X_test`` and ``y_test``. \"\"\"\n", " score = np.zeros((n_estimators,), dtype=np.float64)\n", " for i, y_pred in enumerate(clf.staged_decision_function(X_test)):\n", " score[i] = clf.loss_(y_test, y_pred)\n", " return score\n", "\n", "\n", "def cv_estimate(n_folds=3):\n", " cv = cross_validation.KFold(n=X_train.shape[0], n_folds=n_folds, random_state=42)\n", " cv_clf = ensemble.GradientBoostingClassifier(**params)\n", " val_scores = np.zeros((n_estimators,), dtype=np.float64)\n", " for train, test in cv:\n", " cv_clf.fit(X_train.iloc[train], y_train.iloc[train])\n", " val_scores += heldout_score(cv_clf, X_train.iloc[test], y_train.iloc[test])\n", " val_scores /= n_folds\n", " return val_scores\n", "\n", "\n", "# Estimate best n_estimator using cross-validation\n", "t0 = time()\n", "cv_score = cv_estimate(3)\n", "print('CV score estimate done in %0.3f[s]' %(time() - t0))\n", "\n", "# Compute best n_estimator for test data\n", "test_score = heldout_score(clf, X_test, y_test)\n", "\n", "# negative cumulative sum of oob improvements\n", "cumsum = -np.cumsum(clf.oob_improvement_)\n", "\n", "# min loss according to OOB\n", "oob_best_iter = x[np.argmin(cumsum)]\n", "\n", "# min loss according to test (normalize such that first loss is 0)\n", "test_score -= test_score[0]\n", "test_best_iter = x[np.argmin(test_score)]\n", "\n", "# min loss according to cv (normalize such that first loss is 0)\n", "cv_score -= cv_score[0]\n", "cv_best_iter = x[np.argmin(cv_score)]" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "As stated in the introduction OOB estimate for Gradient Boosting can be used for computing the optimal number of boosting itarations. It has the advantage of being computed on the fly as a byproduct of the algorithm, but it has the disadvantage of being a pessimistic estimate of the actual loss, especially for high number of trees. In the following picture the OOB loss estimate is plotted against test and CV score. The curves represent the cumulative sum of negative improvement as a function of the number of boosting iterations. The vertical line represent the point were the curve is lower." ] }, { "cell_type": "code", "execution_count": 10, "metadata": { "collapsed": false }, "outputs": [ { "data": { "text/html": [ "\n", "
\n", "\n" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "from bokeh.models import LinearAxis, Range1d\n", "\n", "fig = bk.figure(plot_width=700, plot_height=500)\n", "fig.line(x, cumsum, color=\"blue\", legend=\"OOB loss\")\n", "fig.ray(x=oob_best_iter, y=0, length=0, angle=(np.pi / 2), line_width=1, color=\"blue\")\n", "fig.ray(x=oob_best_iter, y=0, length=0, angle=-(np.pi / 2), line_width=1, color=\"blue\")\n", "fig.text(x=oob_best_iter+100, y=-85, text=[\"OOB best\"], text_font_size='8pt')\n", "\n", "fig.line(x, test_score, color=\"red\", legend=\"Test loss\")\n", "fig.ray(x=test_best_iter, y=0, length=0, angle=(np.pi / 2), line_width=1, color=\"red\")\n", "fig.ray(x=test_best_iter, y=0, length=0, angle=-(np.pi / 2), line_width=1, color=\"red\")\n", "fig.text(x=test_best_iter-900, y=-80, text=[\"Text best\"], text_font_size='8pt')\n", "\n", "fig.line(x, cv_score, color=\"green\", legend=\"CV loss\")\n", "fig.ray(x=cv_best_iter, y=0, length=0, angle=(np.pi / 2), line_width=1, color=\"green\")\n", "fig.ray(x=cv_best_iter, y=0, length=0, angle=-(np.pi / 2), line_width=1, color=\"green\")\n", "fig.text(x=cv_best_iter-900, y=-85, text=[\"CV best\"], text_font_size='8pt')\n", "\n", "fig.xaxis.axis_label = \"number of iterations\"\n", "fig.xaxis.axis_label_text_font_size = '10pt'\n", "fig.yaxis.axis_label = \"normalized loss\"\n", "fig.yaxis.axis_label_text_font_size = '10pt'\n", "\n", "bk.show(fig)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "As we can see from the picture, the OOB score start increasing for a low number of iterations, while CV and Test loss continue decreasing. Use this feature only when performing Cross Validation overhead is unacceptable. " ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## 3 Ensemble Methods comparison\n", "\n", "We can use this example to further compare Random Forest and Gradient Boosting Regression Trees. \n", "\n", "Playing with the parameters of RF and GBRT we can compare their performance regarding both accuracy and speed. As can be seen in order to have the same amount of accuracy, GBRT needs a considerable number of iterations with respect to RF (at least with this particular dataset). With a lower number of trees (as low as 100 suffice) RF achieve the same accuracy (or better in some cases) than GBRT with lots (up to 12000) iterations. Moreover the number of iterations causes GBRT algorithm to run in quite amount of time. Compare the time employed by RF to fit the data with that of GBRT; with the same number of trees/iteration GBRT runs up to 5 time slower than RF (in this example). RF can run faster because they can be paralallelized while GBRT cannot, at least in the implementation of scikit-learn. \n", "\n", "Another implementation of GBRT, namely **XGBoost** (e**X**treme **G**radient **Boost**ing, **XGB**) is a library for parallelized and distributed gradient boosting. It is available on github ([here](https://github.com/dmlc/xgboost)) and can run both on single node multiprocessors or on distributed clusters such as hadoop or yarn. It has wrappers for several languages, including python.\n", "\n", "**REMARK.** Please make sure XGBoost is installed on your machine, note that in this case it was installed from source (at the moment this notebook was made, pip install didn't work).\n", "\n", "Follow the instructions [here](https://github.com/dmlc/xgboost/tree/master/python-package) to perform an up-to-date installation from source (if `pip` doesn't work). \n", "\n", "In practice, clone xgboost repository then:\n", "* To make the python module, type `./build.sh` in the root directory of project\n", "* Make sure you have [setuptools](https://pypi.python.org/pypi/setuptools)\n", "* Install with `python setup.py install` from this directory" ] }, { "cell_type": "code", "execution_count": 12, "metadata": { "collapsed": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "XGBoost fit done in 91.590[s]\n", "Accuracy: 0.9778\n" ] } ], "source": [ "import xgboost as xgb\n", "\n", "params = {'n_estimators': 11000, \n", " 'max_depth': 3, \n", " 'subsample': 0.5,\n", " 'learning_rate': 0.1}\n", "\n", "xgb_model = xgb.XGBClassifier(**params)\n", "\n", "t0 = time()\n", "xgb_model.fit(X_train.values, y_train)\n", "print('XGBoost fit done in %0.3f[s]' %(time() - t0))\n", "xgb_acc = xgb_model.score(X_test.values, y_test)\n", "print(\"Accuracy: {:.4f}\".format(xgb_acc))" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "As can be seen in the last code block, XGBoost achieve the same amount of precision of GBRT but in less than a half of the time. Accuracy is still lower than RF but remember that these comparison are made with a particular dataset and performance is dataset specific. In situation (and dataset) where GBRT performs better than RF it is possible to switch to XGB to lower computational time.\n", "\n", "As mentioned above, there are indeed certain datasets where Gradient Boosting methods perform better than Random Forest. One of the advantages of GBRT over RF is its ability to inherently treat additive functions. We can compare the two methods usind a synthetic dataset built with an additive function by using `make_hastie_10_2` as in the previous notebook. Of course this is a toy example used only to show the properties of the algorithm. " ] }, { "cell_type": "code", "execution_count": 13, "metadata": { "collapsed": false }, "outputs": [], "source": [ "X_hastie, y_hastie = datasets.make_hastie_10_2(n_samples=10000, random_state=42)\n", "labels, y_hastie = np.unique(y_hastie, return_inverse=True)\n", "X_hastie_train, X_hastie_test = X_hastie[:2000], X_hastie[2000:]\n", "y_hastie_train, y_hastie_test = y_hastie[:2000], y_hastie[2000:]" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "In the next block we train a GBRT classifier with `max_leaf_nodes = 4` and $1000$ number of iterations. This setting has the same number of leaves as using a tree of depth $2$, which is the best depth for this dataset as we found out in the previous notebook. Changing the parameter type in this case slightly increases performance, probably because the particular problem is best represented by an unbalanced tree." ] }, { "cell_type": "code", "execution_count": 14, "metadata": { "collapsed": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "F1 score: 0.9274920634920636\n" ] } ], "source": [ "gbrt = ensemble.GradientBoostingClassifier(n_estimators=1000, max_leaf_nodes=4)\n", "gbrt.fit(X_hastie_train, y_hastie_train)\n", "y_gbrt = gbrt.predict(X_hastie_test)\n", "\n", "print(\"F1 score: {0}\".format(metrics.f1_score(y_gbrt, y_hastie_test)))" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Next we train a Random Forest choosing parameters with a grid search to reach the best possible result. " ] }, { "cell_type": "code", "execution_count": 15, "metadata": { "collapsed": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Best parameters\n", "max_depth:\tNone\n", "n_estimators:\t500\n", "max_features:\tauto\n", "F1 score: 0.8755507919495058\n" ] } ], "source": [ "rfc = ensemble.RandomForestClassifier(random_state=42, n_jobs=-1)\n", "hastie_params = {'n_estimators':[100, 500],\n", " 'max_features':['auto', 8],\n", " 'max_depth':[None, 5]}\n", "\n", "grid_hastie = grid_search.GridSearchCV(rfc, hastie_params, cv=10, n_jobs=-1)\n", "grid_hastie.fit(X_hastie_train, y_hastie_train)\n", "rfc_best = grid_hastie.best_estimator_\n", "\n", "print(\"Best parameters\")\n", "for k, value in grid_hastie.best_params_.items():\n", " print(k + \":\\t\" + str(value))\n", "\n", "y_rfc = rfc_best.predict(X_hastie_test)\n", "\n", "print(\"F1 score: {0}\".format(metrics.f1_score(y_rfc, y_hastie_test)))" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Comparing score of the two algorithms we can see that GBRT performs better than RF (even if we tuned it a little bit more). This behavior can be explained by the particular dataset used. As noted in the introduction the dataset is generated by an additive function and RF are known to be \"non-additive\" ensemble methods. Note that changing the depth of the trees of the Random Forest to a lower parameter (shallow trees) does not improve performance.\n", "\n", "This behavior could be explained by noting that, in Random Forest, when features exhibit a pure interaction, there is no evident main effect and the first split is essentially a random split (each splitting point does not produce a clear decrese in node impurity). For such reason the algorithm may get lost in the way of building each tree and never really find a suitable function representation." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### 3.1 Ensemble Explore\n", "\n", "In the next image we introduce a software for visualizing Random Forests. It tries to capture the aggregate of all Trees in the forest and to show them in a meaningful way.\n", "\n", "\n", "\n", "The software creates an aggregate tree, working by levels. At each level it counts the frequency of occurrence of each feature in the trees. The aggregate is represented by partitioning a series of concentric rings. Each sector of a ring has amplitude proportional to the number of nodes that belong to the feature. The features are represented by colors. In this way a feature selected frequently at a certain level it will have a larger sector.\n", "\n", "The side panel presents few additional functionalities. For example the color code of the features or a text box that allows the user to interact with the graphics using a special language or the distribution of samples in the leaves. For further details please visit github ([https://github.com/sebastiano-barrera/ensemble-explore](https://github.com/sebastiano-barrera/ensemble-explore)) and download the software.\n", "\n", "As can be seen in the graph, the difficulty of Random Forest to find a suitable model of the dataset, can be spotted by noting that all the features are selected at all levels. There is not a predominant feature that allows the algorithm to find the best split in order to reduce variance.\n", "\n", "In the next paragraph we will compare this graph with the one obtained by a Random Forest on another dataset." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## 4 Feature importance evaluation\n", "\n", "Ensemble methods can provide a measure of importance for each feature in the dataset. This measure can be used to asses how much the feature is important in predicting the response and thus to select the most important features and reduce dimensionality. \n", "\n", "Random Forests have two ways to assign a measure of importance, namely: mean decrease in node impurity and mean decrease in accuracy.\n", "\n", "- _mean decrease in node impurity:_ In scikit-learn the feature importance is provided by looking at each split in each tree. The importance of the splitting variable is proportional to the improvement to the gini index given by that split and it is accumulated (for each variable) over all the trees in the forest.\n", "- _mean decrease in accuracy:_ This method, proposed in the original paper, uses the OOB samples to construct the measure. In practice when a given tree is constructed the OOB samples are passed down the tree and the prediction accuracy is recorded. Then a variable is chosen and its values are permuted (in the OOB samples) and the accuracy is computed again. A decrease in accuracy obtained by this permutation is averaged over all trees for each variable and it provides the importance of that variable (the higher the decreas the higher the importance). This method has better statistical properties with respect to the other and it is used in other implementation such as the package randomForest in the R language. We propose a module that calculate this score for scikit-learn.\n", "\n", "In this example we use two different `ensemble` algorithms on the *Boston Housing Dataset*: first we use a `RandomForestRegressor` algotithm, then we run a `GradientBoostingRegressor` on the same dataset, in order to understand how variable importance differs in the two algorithms.\n", "\n", "Let's start by reading the *Boston Housing Dataset* and storing it properly in a `pandas DataFrame`:" ] }, { "cell_type": "code", "execution_count": 16, "metadata": { "collapsed": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "(506, 13)\n" ] } ], "source": [ "from sklearn.datasets import load_boston\n", "\n", "warnings.filterwarnings(\"ignore\", category=DeprecationWarning)\n", "data = load_boston()\n", "boston = pd.DataFrame(data.data, columns=data.feature_names)#[:-1])\n", "# Assign more descriptive names, see data.DESCR for more info\n", "boston = boston.rename(columns={u'CRIM':'Per Capita Crime Rate',\n", " u'ZN':'Land_zn / lots_over_25k_sqft',\n", " u'INDUS':'Prop non-retail business acres',\n", " u'CHAS':'Bounds Charles River',\n", " u'NOX':'NOX Concentration',\n", " u'RM':'Avg Rooms per Dwelling',\n", " u'AGE':'Prop units built prior 1940',\n", " u'DIS':'Distance from Biz Centers',\n", " u'RAD':'Accessibility to Highways',\n", " u'TAX':'Property TAX rate per $10k',\n", " u'PTRATIO':'Pupil-Teacher Ratio',\n", " u'B':'Prop of Blacks',\n", " u'LSTAT':'% Lower pop Status'})\n", "print(boston.shape)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Split the data in **Training and Validation Set**:" ] }, { "cell_type": "code", "execution_count": 17, "metadata": { "collapsed": false }, "outputs": [], "source": [ "idx_train, idx_valid = cross_validation.train_test_split(boston.index, test_size=0.20)\n", "boston_train, boston_valid = boston.ix[idx_train], boston.ix[idx_valid]\n", "bostony_train, bostony_valid = data.target[idx_train], data.target[idx_valid]" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### 4.1 Random Forest Importance\n", "\n", "In this example we use a Random Forest Regressor and we choose the best estimator based on Cross Validation score." ] }, { "cell_type": "code", "execution_count": 18, "metadata": { "collapsed": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Done in 35.836[s]\n" ] } ], "source": [ "param_grid = [{'n_estimators': [25, 50, 100, 200],\n", " 'max_depth': [16, 32],\n", " 'min_samples_split' : [2, 4],\n", " 'min_samples_leaf' : [1],\n", " 'bootstrap' : [True]}]\n", "t0 = time()\n", "rfr = grid_search.GridSearchCV(ensemble.RandomForestRegressor(),\n", " param_grid, cv=5, verbose=0)\n", "rfr.fit(boston_train, bostony_train)\n", "print('Done in %0.3f[s]' %(time() - t0))" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Next we highlight the best estimator parameters and its outcome." ] }, { "cell_type": "code", "execution_count": 19, "metadata": { "collapsed": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "max_depth : 32\n", "bootstrap : True\n", "min_samples_split : 2\n", "n_estimators : 100\n", "min_samples_leaf : 1\n" ] }, { "data": { "text/html": [ "\n", "
\n", "\n" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "best_estimator = rfr.best_estimator_\n", "for key, value in rfr.best_params_.items():\n", " print(key.ljust(20), ':', value)\n", "\n", "bostony_predic = best_estimator.predict(boston_valid)\n", "\n", "fig = bk.figure(plot_width=500, plot_height=350)\n", "fig.title_text_font_size = '11pt'\n", "fig.circle(bostony_valid, bostony_predic, size=5)\n", "\n", "fig.title = \"True vs Predicted Price\"\n", "fig.xaxis.axis_label = 'True price ($1000s)'\n", "fig.yaxis.axis_label = 'Predicted price ($1000s)'\n", "fig.axis.axis_label_text_font_size = '9pt'\n", "fig.grid.grid_line_color = None\n", "fig.axis.minor_tick_out = 0\n", "fig.axis.major_tick_out = 0\n", "fig.axis.major_tick_in = 3\n", "\n", "bk.show(fig)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We display the features ordered from the most important to the least important. In the next plot we can see the same information but with the bar that corresponds to the relative importance of each feature compared to the most important one. The importance in this case is calculated with the \"gini\" (__mean decrease impurity__) of scikit-learn." ] }, { "cell_type": "code", "execution_count": 20, "metadata": { "collapsed": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Feature ranking:\n", "1. feature Avg Rooms per Dwelling (0.445189)\n", "2. feature % Lower pop Status (0.388411)\n", "3. feature Distance from Biz Centers (0.041747)\n", "4. feature Per Capita Crime Rate (0.038614)\n", "5. feature NOX Concentration (0.021149)\n", "6. feature Property TAX rate per $10k (0.016147)\n", "7. feature Pupil-Teacher Ratio (0.014963)\n", "8. feature Prop units built prior 1940 (0.011141)\n", "9. feature Prop of Blacks (0.009772)\n", "10. feature Prop non-retail business acres (0.007283)\n", "11. feature Accessibility to Highways (0.003056)\n", "12. feature Land_zn / lots_over_25k_sqft (0.001331)\n", "13. feature Bounds Charles River (0.001197)\n" ] } ], "source": [ "feature_importance_raw = best_estimator.feature_importances_\n", "feature_importance = 100.0*(feature_importance_raw/feature_importance_raw.max())\n", "feature_importance_sorted_idx = np.argsort(feature_importance)[::-1]\n", "#print feature_importance_sorted_idx\n", "# Print the feature ranking\n", "print(\"Feature ranking:\")\n", "\n", "for f in range(len(feature_importance_sorted_idx)):\n", " print(\"%d. feature %s (%f)\" % (f + 1, \n", " boston.columns[feature_importance_sorted_idx[f]], \n", " feature_importance_raw[feature_importance_sorted_idx[f]]))" ] }, { "cell_type": "code", "execution_count": 21, "metadata": { "collapsed": false }, "outputs": [ { "data": { "text/html": [ "\n", "
\n", "\n" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "sorted_idx = feature_importance_sorted_idx[::-1]\n", "cols = list(boston.columns[sorted_idx])\n", "fig = bk.figure(plot_width=750, plot_height=400, \n", " title='Variable Importance',\n", " y_range=cols, x_range=(0, 100))\n", "fig.title_text_font_size = '10.5pt'\n", "fig.segment(0, cols, \n", " feature_importance[sorted_idx], cols,\n", " line_width=20)\n", "fig.xaxis.axis_label = 'Relative Importance'\n", "fig.xaxis.axis_label_text_font_size = '9pt'\n", "\n", "bk.show(fig)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The following two cells demonstrate the use of the other methods for calculating feature importance (__mean decrease in accuracy__). We use a module called `importance` provided with `addutils` package." ] }, { "cell_type": "code", "execution_count": 22, "metadata": { "collapsed": false }, "outputs": [], "source": [ "#sys.path.append(os.getcwd() + \"/utilities\")\n", "#import importance\n", "from addutils import importance" ] }, { "cell_type": "code", "execution_count": 23, "metadata": { "collapsed": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Feature ranking:\n", "1. feature % Lower pop Status (60.749022)\n", "2. feature Avg Rooms per Dwelling (47.871714)\n", "3. feature Per Capita Crime Rate (6.897790)\n", "4. feature NOX Concentration (5.163566)\n", "5. feature Distance from Biz Centers (5.020479)\n", "6. feature Property TAX rate per $10k (2.850435)\n", "7. feature Pupil-Teacher Ratio (2.549015)\n", "8. feature Prop units built prior 1940 (1.982671)\n", "9. feature Prop non-retail business acres (1.785406)\n", "10. feature Accessibility to Highways (0.879166)\n", "11. feature Prop of Blacks (0.679599)\n", "12. feature Land_zn / lots_over_25k_sqft (0.072743)\n", "13. feature Bounds Charles River (0.008772)\n" ] } ], "source": [ "mda_feature_importance_raw = importance.importance(best_estimator, \n", " boston_train.values, \n", " bostony_train, \n", " importance.reg_score)\n", "mda_feature_importance_raw = mda_feature_importance_raw.mean(axis=0)\n", "mda_feature_importance = 100.0*(mda_feature_importance_raw/mda_feature_importance_raw.max())\n", "mda_feature_importance_sorted_idx = np.argsort(mda_feature_importance)[::-1]\n", "#print feature_importance_sorted_idx\n", "# Print the feature ranking\n", "print(\"Feature ranking:\")\n", "\n", "for f in range(len(mda_feature_importance_sorted_idx)):\n", " print(\"%d. feature %s (%f)\" % (f + 1, \n", " boston.columns[mda_feature_importance_sorted_idx[f]], \n", " mda_feature_importance_raw[mda_feature_importance_sorted_idx[f]]))" ] }, { "cell_type": "code", "execution_count": 24, "metadata": { "collapsed": false, "scrolled": false }, "outputs": [ { "data": { "text/html": [ "\n", "
\n", "\n" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "mda_sorted_idx = mda_feature_importance_sorted_idx[::-1]\n", "cols = list(boston.columns[mda_sorted_idx])\n", "fig = bk.figure(plot_width=750, plot_height=400, \n", " title='Variable Importance',\n", " y_range=cols, x_range=(0, 100))\n", "fig.title_text_font_size = '10.5pt'\n", "fig.segment(0, cols, \n", " mda_feature_importance[mda_sorted_idx], cols,\n", " line_width=20)\n", "fig.xaxis.axis_label = 'Relative Importance'\n", "fig.xaxis.axis_label_text_font_size = '9pt'\n", "\n", "bk.show(fig)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "In the literature the two approaches differs mainly because _gini impurity_ is biased in favor of continuous variables and variables with many categories (for variables of different types), while _mean decrease accuracy_ is unbiased (only when subsampling is not used).\n", "\n", "As long as this dataset is concerned, the two approaches have similar results. This can be explained by the fact that all variables are of the same type and thus either approach can be used succesfully. " ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "#### 4.1.1 Ensemble Explore\n", "The next picture shows the representation of this Random Forest with aggregate trees, intoduced above. \n", "\n", "\n", "\n", "As can be seen the most important features *Avg Rooms per Dwelling* (RM) and *% Lower pop Status* (LSTAT) are selected as the root and second level by the majority of the trees. Moreover many trees have path that include those two features, from the root to the leaves." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### 4.2 Gradient Boosting Importance\n", "\n", "In this example we use a Gradient Boosting Regressor Tree and we choose the best estimator based on Cross Validation score." ] }, { "cell_type": "code", "execution_count": 25, "metadata": { "collapsed": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Done in 6.332[s]\n" ] } ], "source": [ "gbr = ensemble.GradientBoostingRegressor()\n", "params = {'n_estimators':[100, 200, 500],\n", " 'max_depth':[4, 6],\n", " 'learning_rate':[0.1, 0.01],\n", " 'subsample':[0.5]}\n", "t0 = time()\n", "grid = grid_search.GridSearchCV(gbr, params, n_jobs=-1)\n", "grid.fit(boston_train, bostony_train)\n", "\n", "print('Done in %0.3f[s]' %(time() - t0))" ] }, { "cell_type": "code", "execution_count": 26, "metadata": { "collapsed": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "max_depth : 6\n", "learning_rate : 0.1\n", "n_estimators : 100\n", "subsample : 0.5\n" ] } ], "source": [ "gbr_best = grid.best_estimator_\n", "for key, value in grid.best_params_.items():\n", " print(key.ljust(20), ':', value)\n", "\n", "bostony_predic = gbr_best.predict(boston_valid)" ] }, { "cell_type": "code", "execution_count": 27, "metadata": { "collapsed": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Feature ranking:\n", "1. feature Avg Rooms per Dwelling (0.191616)\n", "2. feature % Lower pop Status (0.182176)\n", "3. feature Prop of Blacks (0.120345)\n", "4. feature Distance from Biz Centers (0.105044)\n", "5. feature Prop units built prior 1940 (0.101373)\n", "6. feature Per Capita Crime Rate (0.092349)\n", "7. feature NOX Concentration (0.059848)\n", "8. feature Property TAX rate per $10k (0.042669)\n", "9. feature Pupil-Teacher Ratio (0.030711)\n", "10. feature Prop non-retail business acres (0.028295)\n", "11. feature Accessibility to Highways (0.022802)\n", "12. feature Land_zn / lots_over_25k_sqft (0.013333)\n", "13. feature Bounds Charles River (0.009439)\n" ] } ], "source": [ "gbr_feature_importance_raw = gbr_best.feature_importances_\n", "gbr_feature_importance = 100.0*(gbr_feature_importance_raw/gbr_feature_importance_raw.max())\n", "gbr_feature_importance_sorted_idx = np.argsort(gbr_feature_importance)[::-1]\n", "\n", "print(\"Feature ranking:\")\n", "\n", "for f in range(len(feature_importance_sorted_idx)):\n", " print(\"%d. feature %s (%f)\" % (f + 1, \n", " boston.columns[gbr_feature_importance_sorted_idx[f]], \n", " gbr_feature_importance_raw[gbr_feature_importance_sorted_idx[f]]))" ] }, { "cell_type": "code", "execution_count": 28, "metadata": { "collapsed": false, "scrolled": false }, "outputs": [ { "data": { "text/html": [ "\n", "
\n", "\n" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "gbr_sorted_idx = gbr_feature_importance_sorted_idx[::-1]\n", "cols = list(boston.columns[gbr_sorted_idx])\n", "fig = bk.figure(plot_width=700, plot_height=400,\n", " title='Variable Importance',\n", " y_range=cols, x_range=(0, 100))\n", "fig.title_text_font_size = '10.5pt'\n", "fig.segment(0, cols, \n", " gbr_feature_importance[gbr_sorted_idx], cols,\n", " line_width=20)\n", "fig.xaxis.axis_label = 'Relative Importance'\n", "fig.xaxis.axis_label_text_font_size = '9pt'\n", "\n", "bk.show(fig)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Due to shrinkage, GBRT variable importance has the advantage, over Random Forest, to be less susceptible to the masking effect that high correlated variables have; that is it should be easier to spot all important variables even if they are correlated." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## 5 Partial Dependence Plots\n", "**Partial dependence plots (PDP)** show the dependence between the target response and a set of ‘target’ features, marginalizing over the values of all other features (the ‘complement’ features). Intuitively, we can interpret the partial dependence as the expected target response as a function of the ‘target’ features.\n", "\n", "Due to the limits of human perception the size of the target feature set must be small (usually, one or two) thus the target features are usually chosen among the most important features. In the next section we evaluate the importance of features using Random Forest and then we use only the most important ones in the plot.\n", "\n", "### 5.1 California Housing Dataset\n", "\n", "California Housing is obtained from the StatLib repository. Here is the included description:\n", "\n", "*S&P Letters Data\n", "We collected information on the variables using all the block groups in California from the 1990 Cens us. In this sample a block group on average includes 1425.5 individuals living in a geographically co mpact area. Naturally, the geographical area included varies inversely with the population density. We computed distances among the centroids of each block group as measured in latitude and longitude. We excluded all the block groups reporting zero entries for the independent and dependent variables. The final data contained 20,640 observations on 9 variables. The dependent variable is ln(median house value).*\n", "\n" ] }, { "cell_type": "code", "execution_count": 29, "metadata": { "collapsed": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "downloading Cal. housing from http://lib.stat.cmu.edu/modules.php?op=modload&name=Downloads&file=index&req=getit&lid=83 to temp/\n" ] } ], "source": [ "from sklearn.datasets.california_housing import fetch_california_housing\n", "\n", "# fetch California housing dataset\n", "cal_housing = fetch_california_housing(\"temp/\")\n", "\n", "Cal_train, Cal_test, Caly_train, Caly_test = cross_validation.train_test_split(cal_housing.data, \n", " cal_housing.target,\n", " test_size=0.2,\n", " random_state=1)\n", "names = cal_housing.feature_names" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "In the following example, we use scikit-learn's RandomForestRegressors to asses variable importance. " ] }, { "cell_type": "code", "execution_count": 30, "metadata": { "collapsed": false }, "outputs": [ { "data": { "text/plain": [ "RandomForestRegressor(bootstrap=True, criterion='mse', max_depth=None,\n", " max_features=0.3, max_leaf_nodes=None, min_samples_leaf=1,\n", " min_samples_split=2, min_weight_fraction_leaf=0.0,\n", " n_estimators=500, n_jobs=1, oob_score=False, random_state=None,\n", " verbose=0, warm_start=False)" ] }, "execution_count": 30, "metadata": {}, "output_type": "execute_result" } ], "source": [ "forest = ensemble.RandomForestRegressor(n_estimators=500, max_features=0.3)\n", "forest.fit(Cal_train, Caly_train)" ] }, { "cell_type": "code", "execution_count": 31, "metadata": { "collapsed": false }, "outputs": [], "source": [ "importances_raw = forest.feature_importances_\n", "importances = 100.0 * (importances_raw / importances_raw.max())\n", "#std = np.std([tree.feature_importances_ for tree in forest.estimators_], axis=0)\n", "indices = np.argsort(importances)[::-1]" ] }, { "cell_type": "code", "execution_count": 32, "metadata": { "collapsed": false }, "outputs": [ { "data": { "text/html": [ "\n", "
\n", "\n" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "rev_indices = indices[::-1]\n", "cols = [names[r] for r in rev_indices]\n", "fig = bk.figure(plot_width=700, plot_height=300,\n", " title='Variable Importance',\n", " y_range=cols, x_range=(0, 100))\n", "fig.title_text_font_size = '10.5pt'\n", "fig.segment(0, \n", " cols, \n", " importances[rev_indices], \n", " cols,\n", " line_width=20)\n", "fig.xaxis.axis_label = 'Relative Importance'\n", "fig.xaxis.axis_label_text_font_size = '9pt'\n", "\n", "bk.show(fig)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Here the length of the bar represents the percentage value of feature importance relative to the most important feature. The plot suggests that one features, namely MedInc captures most of the variability in the data.\n", "\n", "In the following plot we can see variable importance calculated with the method \"mean decrease accuracy\"." ] }, { "cell_type": "code", "execution_count": 33, "metadata": { "collapsed": false }, "outputs": [], "source": [ "mda_importance = importance.importance(forest, Cal_train, Caly_train, importance.reg_score)\n", "mda_importances_raw = mda_importance.mean(axis=0)\n", "mda_importances = 100.0 * (mda_importances_raw / mda_importances_raw.max())\n", "mda_indices = np.argsort(mda_importances)[::-1]" ] }, { "cell_type": "code", "execution_count": 34, "metadata": { "collapsed": false, "scrolled": true }, "outputs": [ { "data": { "text/html": [ "\n", "
\n", "\n" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "rev_mda_indices = mda_indices[::-1]\n", "cols = [names[r] for r in rev_mda_indices]\n", "fig = bk.figure(plot_width=700, plot_height=300,\n", " title='Variable Importance',\n", " y_range=cols, x_range=(0, 100))\n", "fig.title_text_font_size = '10.5pt'\n", "fig.segment(0, \n", " cols, \n", " mda_importances[rev_mda_indices], \n", " cols,\n", " line_width=20)\n", "fig.xaxis.axis_label = 'Relative Importance'\n", "fig.xaxis.axis_label_text_font_size = '9pt'\n", "\n", "bk.show(fig)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Both measures says that Median Income in the neighborhood is the most relevant feature (not surprisingly). Longitude, Latitude, and Average Occupancy are still influencial and we choose to include HouseAge and AveRooms because they can give more insights into the driving factor of house price. With only latitude and longitude features, the model learns the price of the house given its position regardless of other factor that may affect the price." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### 5.2 Partial Dependence Plots\n", "\n", "Based on the observation above we plot only the first most important features. In the picture below the plot shows six single-variable partial dependence. The plots are not smooth because we are using a tree based method. The marks at the bottom of each plots delineates the deciles of the data distribution of the variable. Note that the density is lower near the edges." ] }, { "cell_type": "code", "execution_count": 35, "metadata": { "collapsed": false }, "outputs": [], "source": [ "from sklearn.cross_validation import train_test_split\n", "from sklearn.ensemble.partial_dependence import plot_partial_dependence\n", "from sklearn.ensemble.partial_dependence import partial_dependence" ] }, { "cell_type": "code", "execution_count": 36, "metadata": { "collapsed": false }, "outputs": [ { "data": { "text/plain": [ "GradientBoostingRegressor(alpha=0.9, init=None, learning_rate=0.1,\n", " loss='huber', max_depth=4, max_features=None,\n", " max_leaf_nodes=None, min_samples_leaf=1, min_samples_split=2,\n", " min_weight_fraction_leaf=0.0, n_estimators=100,\n", " random_state=1, subsample=1.0, verbose=0, warm_start=False)" ] }, "execution_count": 36, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf = ensemble.GradientBoostingRegressor(n_estimators=100, max_depth=4,\n", " learning_rate=0.1, loss='huber', random_state=1)\n", "clf.fit(Cal_train, Caly_train)" ] }, { "cell_type": "code", "execution_count": 37, "metadata": { "collapsed": true }, "outputs": [], "source": [ "warnings.filterwarnings('ignore', category=UnicodeWarning)\n", "warnings.filterwarnings('ignore', category=FutureWarning)" ] }, { "cell_type": "code", "execution_count": 38, "metadata": { "collapsed": false }, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAyIAAAL4CAYAAACHn0OjAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/0U9eZ+LHOqiAAAAAElFTkSuQmCC\n", "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# sklearn uses Matplotlib for its plotting functions (like `plot_partial_dependence')\n", "%matplotlib inline\n", "import matplotlib.pyplot as plt \n", "\n", "plt_number = 6\n", "features = list(indices[:plt_number])\n", "#features += zip(features[0::2], features[1::2])\n", "fig = plt.figure(figsize=(12,12))\n", "ax = fig.add_subplot(111)\n", "\n", "fig, axs = plot_partial_dependence(clf, Cal_train, features, feature_names=names,\n", " n_jobs=-1, grid_resolution=100, ax=ax)\n", "fig.suptitle('Partial dependence of house value w.r.t. most important features\\n'\n", " 'for the California housing dataset')\n", "plt.subplots_adjust(top=0.9) # tight_layout causes overlap with suptitle" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The figure above shows four one-way and two two-way partial dependence plots for the California housing dataset: the target variables for the one-way PDP are: median income (MedInc), avg. occupants per household (AvgOccup), latitude (Latitude), longitude (Longitude).\n", "\n", "The partial dependence of median house price shows a (monotonically increasing) linear relationship with the median income (top left). House value is generally monotonically decreasing with increasing average occupancy (top middle). \n", "\n", "Median house value has a weak partial dependence with HouseAge that is inconsistent with its ranking (and common sense). This behavior may suggest an interaction effect with other variables. The following plot shows a two-variable dependency plot of house value on joint values of house age and average occupancy" ] }, { "cell_type": "code", "execution_count": 39, "metadata": { "collapsed": false }, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaQAAAEuCAYAAADBZ2aLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n", "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "from mpl_toolkits.mplot3d import Axes3D\n", "\n", "fig = plt.figure()\n", "\n", "#target_feature = list(indices[:2])\n", "target_feature = (1,5)\n", "pdp, (x_axis, y_axis) = partial_dependence(clf, target_feature,\n", " X=Cal_train, grid_resolution=50)\n", "XX, YY = np.meshgrid(x_axis, y_axis)\n", "Z = pdp.T.reshape(XX.shape).T\n", "ax = Axes3D(fig)\n", "surf = ax.plot_surface(XX, YY, Z, rstride=1, cstride=1, cmap=plt.cm.BuPu)\n", "ax.set_xlabel(names[target_feature[0]])\n", "ax.set_ylabel(names[target_feature[1]])\n", "ax.set_zlabel('Partial dependence')\n", "# pretty init view\n", "ax.view_init(elev=22, azim=122)\n", "plt.colorbar(surf)\n", "plt.suptitle('Partial dependence of house value on \\n'\n", " 'median age and average occupancy')\n", "plt.subplots_adjust(top=0.9)\n", "\n", "plt.show()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The two-way partial dependence plot shows the dependence of median house price on joint values of house age and avg. occupants per household. We can clearly see an interaction between the two features: for an avg. occupancy greater than two, the house price is nearly independent of the house age, whereas for values less than two there is a strong dependence on age." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## 6 Ensemble Methods, practical considerations\n", "\n", "Ensemble methods are non-parametric models with few hyperparameters to set. They can model complex iteraction between variable and learn complex functions. In the next section we will try to sketch a possible use cases for Ensemble Methods.\n", "\n", "- Is there an a-priori knowledge of the model underlying the particular dataset?\n", " 1. NO. use a non-parametric model such as RF or GBRT\n", " 2. YES. use a parametric model (such as linear regression). It is still usefull though to compare results of a parametric model with results of a RF or GBRT especially if the former does not achieve good performance. \n", "\n", "- Do we know that the model is sparse?\n", " 1. YES. Use a sparse model such as LASSO. Random Forest performance in this case are worse. Lasso however doesn't automatically capture high order correlation between variables and non-linear relationship. All the interaction and nonlinearties must be explicitly added by the user. Doing this could be hard because the user need to check many possible combination of features and nonlinear contribution (square, cube, etc.). If the model is sparse but it is still difficoult to manage all possible feature combinations it is worth trying to use Support Vector Machines that automatically detect non linear models.\n", " 2. NO. If feature importance of a Random Forest displays high variability it might be the that few features are relevant and the algorithm cannot capture that relationship. In presence of such behavior it could be worth trying with LASSO or other sparse methods.\n", " \n", "- Is the model known to be additive?\n", " 1. YES. Try using GBRT instead of RF. If performance are an issue, use XGB. Compare results with **Generalized Additive Models** (**GAM**).\n", " 2. NO. Try using a RF and then GBRT. Compare results to check if the performance differ. Try spotting additive model with visualization. If the model is additive try to improve it with GAM (it has a smoother function w.r.t. ensembles)\n", " \n", "- Is the dataset Big?\n", " 1. YES. Ensemble methods, especially RF, scale and parallelize well. Given their many properties, if the dataset is big and we are analyzing the data for the first time, Ensamble Methods are a good first choice. They are also descriptive and can give further hints on the data at hand, for example variable importance." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "---\n", "\n", "Visit [www.add-for.com]() for more tutorials and updates.\n", "\n", "This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License." ] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.4.3" } }, "nbformat": 4, "nbformat_minor": 0 }