Test case for Isotonic Regression regression in fit vs. fit_transform

isotonic_test_case_20150129.json

{"nbformat_minor": 0, "cells": [{"execution_count": 1, "cell_type": "code", "source": "# Imports\nimport matplotlib.pyplot as plt\nimport numpy\nimport pandas\nimport scipy\nimport sklearn\nimport sklearn.isotonic\nimport sys", "outputs": [], "metadata": {"collapsed": true, "trusted": true}}, {"source": "## Version strings", "cell_type": "markdown", "metadata": {}}, {"execution_count": 2, "cell_type": "code", "source": "print(sys.version)\nprint(sklearn.__version__)", "outputs": [{"output_type": "stream", "name": "stdout", "text": "2.7.3 (default, Mar 13 2014, 11:03:55) \n[GCC 4.7.2]\n0.16.dev\n"}], "metadata": {"collapsed": false, "trusted": true}}, {"source": "## Generate samples with and without ties", "cell_type": "markdown", "metadata": {}}, {"execution_count": 3, "cell_type": "code", "source": "# Sample with x ties\ndata_with_ties = pandas.DataFrame()\ndata_with_ties[\"feature\"] = [0, 0, 1, 2, 3]\ndata_with_ties[\"target\"] = [0.1, 0.05, 0.15, 0.2, 0.35]\n\n# Sample without x ties\ndata_without_ties = pandas.DataFrame()\ndata_without_ties[\"feature\"] = [0, 1, 2, 3, 4]\ndata_without_ties[\"target\"] = [0.1, 0.05, 0.15, 0.2, 0.35]\n", "outputs": [], "metadata": {"collapsed": false, "trusted": true}}, {"source": "## Failure test case with ties", "cell_type": "markdown", "metadata": {}}, {"execution_count": 4, "cell_type": "code", "source": "%matplotlib inline\n\n# Fit and then transform the model\nisotonic = sklearn.isotonic.IsotonicRegression()\nisotonic.fit(data_with_ties.loc[:, \"feature\"],\n            data_with_ties.loc[:, \"target\"])\nresults_fit = isotonic.transform(data_with_ties.loc[:, \"feature\"])\n\n# fit_transform in one go\nisotonic = sklearn.isotonic.IsotonicRegression()\nresults_fit_transform = isotonic.fit_transform(data_with_ties.loc[:, \"feature\"],\n             data_with_ties.loc[:, \"target\"])\n\n# Describe data and output norm\nprint(\"Norm: {0}\".format(numpy.linalg.norm(results_fit - results_fit_transform)))\nprint(\"fit, transform\")\nprint(pandas.DataFrame(results_fit).describe())\nprint(\"fit_transform\")\nprint(pandas.DataFrame(results_fit_transform).describe())\n\n# Plot\nf = plt.figure()\nplt.scatter(data_with_ties.loc[:, \"feature\"],\n            data_with_ties.loc[:, \"target\"],\n            color=\"red\", alpha=0.25)\nplt.scatter(data_with_ties.loc[:, \"feature\"],\n            results_fit,\n            color=\"green\", alpha=0.25)\nplt.scatter(data_with_ties.loc[:, \"feature\"],\n            results_fit_transform,\n            color=\"blue\", alpha=0.25)\n_ = plt.legend((\"Actual\", \"fit, transform\", \"fit_transform\"),\n              loc=\"best\")", "outputs": [{"output_type": "stream", "name": "stdout", "text": "Norm: 0.111803398875\nfit, transform\n              0\ncount  5.000000\nmean   0.140000\nstd    0.147479\nmin    0.000000\n25%    0.000000\n50%    0.150000\n75%    0.200000\nmax    0.350000\nfit_transform\n              0\ncount  5.000000\nmean   0.170000\nstd    0.115109\nmin    0.050000\n25%    0.100000\n50%    0.150000\n75%    0.200000\nmax    0.350000\n"}, {"output_type": "display_data", "data": {"image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEACAYAAABcXmojAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n", "text/plain": "<matplotlib.figure.Figure at 0x46dd9d0>"}, "metadata": {}}], "metadata": {"collapsed": false, "trusted": true}}, {"source": "## Success case without ties", "cell_type": "markdown", "metadata": {}}, {"execution_count": 5, "cell_type": "code", "source": "%matplotlib inline\n\n# Fit and then transform the model\nisotonic = sklearn.isotonic.IsotonicRegression()\nisotonic.fit(data_without_ties.loc[:, \"feature\"],\n            data_without_ties.loc[:, \"target\"])\nresults_fit = isotonic.transform(data_without_ties.loc[:, \"feature\"])\n\n# fit_transform in one go\nisotonic = sklearn.isotonic.IsotonicRegression()\nresults_fit_transform = isotonic.fit_transform(data_without_ties.loc[:, \"feature\"],\n             data_without_ties.loc[:, \"target\"])\n\n# Describe data and output norm\nprint(\"Norm: {0}\".format(numpy.linalg.norm(results_fit - results_fit_transform)))\nprint(\"fit, transform\")\nprint(pandas.DataFrame(results_fit).describe())\nprint(\"fit_transform\")\nprint(pandas.DataFrame(results_fit_transform).describe())\n\n# Plot\nf = plt.figure()\nplt.scatter(data_without_ties.loc[:, \"feature\"],\n            data_without_ties.loc[:, \"target\"],\n            color=\"red\", alpha=0.25)\nplt.scatter(data_without_ties.loc[:, \"feature\"],\n            results_fit,\n            color=\"green\", alpha=0.25)\nplt.scatter(data_without_ties.loc[:, \"feature\"],\n            results_fit_transform,\n            color=\"blue\", alpha=0.25)\n_ = plt.legend((\"Actual\", \"fit, transform\", \"fit_transform\"),\n              loc=\"best\")", "outputs": [{"output_type": "stream", "name": "stdout", "text": "Norm: 0.0\nfit, transform\n              0\ncount  5.000000\nmean   0.170000\nstd    0.113743\nmin    0.075000\n25%    0.075000\n50%    0.150000\n75%    0.200000\nmax    0.350000\nfit_transform\n              0\ncount  5.000000\nmean   0.170000\nstd    0.113743\nmin    0.075000\n25%    0.075000\n50%    0.150000\n75%    0.200000\nmax    0.350000\n"}, {"output_type": "display_data", "data": {"image/png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAAEACAYAAAC57G0KAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nZ0HVX7kAAAAASUVORK5CYII=\n", "text/plain": "<matplotlib.figure.Figure at 0x478cf10>"}, "metadata": {}}], "metadata": {"collapsed": false, "trusted": true}}], "nbformat": 4, "metadata": {"kernelspec": {"display_name": "Python 2", "name": "python2"}, "language_info": {"mimetype": "text/x-python", "nbconvert_exporter": "python", "version": "2.7.3", "name": "python", "file_extension": ".py", "pygments_lexer": "ipython2", "codemirror_mode": {"version": 2, "name": "ipython"}}}}