{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "# Robust linear regression\n", "### Dr. Tirthajyoti Sarkar, Fremont, CA 94536\n", "\n", "In this notebook, we demonstrate `HuberRegressor` class from scikit-learn which is robust to outliers." ] }, { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "import numpy as np\n", "from sklearn.linear_model import HuberRegressor, LinearRegression\n", "from sklearn.datasets import make_regression\n", "import matplotlib.pyplot as plt" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Creating a regression problem using `make_regression` method" ] }, { "cell_type": "code", "execution_count": 35, "metadata": {}, "outputs": [], "source": [ "rng = np.random.RandomState(0)\n", "\n", "X, y, coef = make_regression(\n", " n_samples=200, n_features=2, noise=4.0, \n", " coef=True, random_state=0\n", ")" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Plot" ] }, { "cell_type": "code", "execution_count": 38, "metadata": {}, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "fix, ax = plt.subplots(1,2,figsize=(10,3))\n", "ax[0].scatter(X[:,0],y)\n", "ax[0].grid(True)\n", "ax[1].scatter(X[:,1],y)\n", "ax[1].grid(True)\n", "plt.show()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Inserting random outliers in the data" ] }, { "cell_type": "code", "execution_count": 39, "metadata": {}, "outputs": [], "source": [ "X[:4] = rng.uniform(10, 20, (4, 2))\n", "y[:4] = rng.uniform(100, 200, 4)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Plot to show the inserted outliers" ] }, { "cell_type": "code", "execution_count": 40, "metadata": {}, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "fix, ax = plt.subplots(1,2,figsize=(10,3))\n", "ax[0].scatter(X[:,0],y)\n", "ax[0].grid(True)\n", "ax[1].scatter(X[:,1],y)\n", "ax[1].grid(True)\n", "plt.show()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Create a `HuberRegressor` object and fit\n", "\n", "The Huber Regressor optimizes the squared loss for the samples where $|\\frac{y - X'w}{\\sigma}| < \\epsilon$ and the absolute loss for the samples where $|\\frac{y - X'w}{\\sigma}| < \\epsilon$, where $w$ and $\\sigma$ are parameters to be optimized. \n", "\n", "The parameter $\\sigma$ makes sure that if $y$ is scaled up or down by a certain factor, one does not need to rescale epsilon to achieve the same robustness. Note that this does not take into account the fact that the different features of $X$ may be of different scales." ] }, { "cell_type": "code", "execution_count": 43, "metadata": {}, "outputs": [], "source": [ "huber = HuberRegressor()" ] }, { "cell_type": "code", "execution_count": 44, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "HuberRegressor(alpha=0.0001, epsilon=1.35, fit_intercept=True, max_iter=100,\n", " tol=1e-05, warm_start=False)" ] }, "execution_count": 44, "metadata": {}, "output_type": "execute_result" } ], "source": [ "huber.fit(X,y)" ] }, { "cell_type": "code", "execution_count": 51, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[16.02763376, 15.44883183]])" ] }, "execution_count": 51, "metadata": {}, "output_type": "execute_result" } ], "source": [ "X[1].reshape(1,-1)" ] }, { "cell_type": "code", "execution_count": 50, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([763.50484115])" ] }, "execution_count": 50, "metadata": {}, "output_type": "execute_result" } ], "source": [ "huber.predict(X[1].reshape(1,-1))" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### A simple linear regression fit for comparison" ] }, { "cell_type": "code", "execution_count": 23, "metadata": {}, "outputs": [], "source": [ "linear = LinearRegression()" ] }, { "cell_type": "code", "execution_count": 24, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "LinearRegression(copy_X=True, fit_intercept=True, n_jobs=None, normalize=False)" ] }, "execution_count": 24, "metadata": {}, "output_type": "execute_result" } ], "source": [ "linear.fit(X, y)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Compare the estimated coefficients\n", "\n", "As we will see, the Huber regressor estimates the coefficients, in the presence of outliers, far better (closer to the true values) than a simple linear model." ] }, { "cell_type": "code", "execution_count": 52, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "True coefficients: [20.4923687 34.16981149]\n", "Huber coefficients: [17.79064226 31.01065329]\n", "Linear Regression coefficients: [ 3.63326689 10.82594129]\n" ] } ], "source": [ "print(\"True coefficients:\", coef)\n", "print(\"Huber coefficients:\", huber.coef_)\n", "print(\"Linear Regression coefficients:\", linear.coef_)" ] }, { "cell_type": "code", "execution_count": 65, "metadata": {}, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "fix, ax = plt.subplots(1,2,figsize=(12,3),sharey=True)\n", "ax[0].barh(['True coef','Huber coef', 'Linear fit coef'],width=[coef[0],huber.coef_[0],linear.coef_[0]])\n", "ax[0].set_title(\"1st coefficients\",fontsize=15)\n", "ax[1].barh(['True coef','Huber coef', 'Linear fit coef'],width=[coef[1],huber.coef_[1],linear.coef_[1]])\n", "ax[1].set_title(\"2nd coefficients\",fontsize=15)\n", "plt.show()" ] } ], "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": "3.6.2" }, "latex_envs": { "LaTeX_envs_menu_present": true, "autoclose": false, "autocomplete": true, "bibliofile": "biblio.bib", "cite_by": "apalike", "current_citInitial": 1, "eqLabelWithNumbers": true, "eqNumInitial": 1, "hotkeys": { "equation": "Ctrl-E", "itemize": "Ctrl-I" }, "labels_anchors": false, "latex_user_defs": false, "report_style_numbering": false, "user_envs_cfg": false } }, "nbformat": 4, "nbformat_minor": 4 }