ContrastiveTests.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Notebook for demonstrating the use of beeswarms for CIU (and other) visualisation."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Contrastive explanations answer questions such as \"Why alternative A rather than B\" or \"Why not alternative B rather than A\". In classification tasks, these questions might rather be of the kind \"Why is this a cat and not a dog?\".\n",
    "\n",
    "CIU can provide answers to such questions becasue any value in the range [0,1] can be used for neutral.CU, including CU values of an instance to compare with that then provide the \"reference value\" to compare against. \n",
    "\n",
    "First some basic imports."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "#%pip install seaborn\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from sklearn.model_selection import train_test_split\n",
    "import xgboost as xgb\n",
    "import ciu as ciu\n",
    "from ciu import ciuplots\n",
    "from ciu.CIU import contrastive_ciu"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Iris data set"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We get LDA model, an instance to explain and the CIU object for them."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from ciu_tests import iris_lda\n",
    "\n",
    "np.random.seed(24)\n",
    "CIU_iris, iris_lda_model, iris_instance = iris_lda.get_iris_test()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>CI</th>\n",
       "      <th>CU</th>\n",
       "      <th>Cinfl</th>\n",
       "      <th>outname</th>\n",
       "      <th>outval</th>\n",
       "      <th>feature</th>\n",
       "      <th>ymin</th>\n",
       "      <th>ymax</th>\n",
       "      <th>inputs</th>\n",
       "      <th>invals</th>\n",
       "      <th>neutralCU</th>\n",
       "      <th>target_concept</th>\n",
       "      <th>target_inputs</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>s_length</th>\n",
       "      <td>0.988743</td>\n",
       "      <td>0.814143</td>\n",
       "      <td>0.310607</td>\n",
       "      <td>versicolor</td>\n",
       "      <td>0.816231</td>\n",
       "      <td>s_length</td>\n",
       "      <td>1.125260e-02</td>\n",
       "      <td>0.999995</td>\n",
       "      <td>[0]</td>\n",
       "      <td>[2.0]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>s_width</th>\n",
       "      <td>0.881502</td>\n",
       "      <td>0.795213</td>\n",
       "      <td>0.260231</td>\n",
       "      <td>versicolor</td>\n",
       "      <td>0.816231</td>\n",
       "      <td>s_width</td>\n",
       "      <td>1.152488e-01</td>\n",
       "      <td>0.996751</td>\n",
       "      <td>[1]</td>\n",
       "      <td>[3.2]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>p_length</th>\n",
       "      <td>0.999546</td>\n",
       "      <td>0.816601</td>\n",
       "      <td>0.316458</td>\n",
       "      <td>versicolor</td>\n",
       "      <td>0.816231</td>\n",
       "      <td>p_length</td>\n",
       "      <td>2.910413e-17</td>\n",
       "      <td>0.999546</td>\n",
       "      <td>[2]</td>\n",
       "      <td>[1.8]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>p_width</th>\n",
       "      <td>0.999996</td>\n",
       "      <td>0.816231</td>\n",
       "      <td>0.316230</td>\n",
       "      <td>versicolor</td>\n",
       "      <td>0.816231</td>\n",
       "      <td>p_width</td>\n",
       "      <td>3.002736e-06</td>\n",
       "      <td>0.999999</td>\n",
       "      <td>[3]</td>\n",
       "      <td>[2.4]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                CI        CU     Cinfl     outname    outval   feature  \\\n",
       "s_length  0.988743  0.814143  0.310607  versicolor  0.816231  s_length   \n",
       "s_width   0.881502  0.795213  0.260231  versicolor  0.816231   s_width   \n",
       "p_length  0.999546  0.816601  0.316458  versicolor  0.816231  p_length   \n",
       "p_width   0.999996  0.816231  0.316230  versicolor  0.816231   p_width   \n",
       "\n",
       "                  ymin      ymax inputs invals  neutralCU target_concept  \\\n",
       "s_length  1.125260e-02  0.999995    [0]  [2.0]        0.5           None   \n",
       "s_width   1.152488e-01  0.996751    [1]  [3.2]        0.5           None   \n",
       "p_length  2.910413e-17  0.999546    [2]  [1.8]        0.5           None   \n",
       "p_width   3.002736e-06  0.999999    [3]  [2.4]        0.5           None   \n",
       "\n",
       "         target_inputs  \n",
       "s_length          None  \n",
       "s_width           None  \n",
       "p_length          None  \n",
       "p_width           None  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>CI</th>\n",
       "      <th>CU</th>\n",
       "      <th>Cinfl</th>\n",
       "      <th>outname</th>\n",
       "      <th>outval</th>\n",
       "      <th>feature</th>\n",
       "      <th>ymin</th>\n",
       "      <th>ymax</th>\n",
       "      <th>inputs</th>\n",
       "      <th>invals</th>\n",
       "      <th>neutralCU</th>\n",
       "      <th>target_concept</th>\n",
       "      <th>target_inputs</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>s_length</th>\n",
       "      <td>0.183769</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.091885</td>\n",
       "      <td>virginica</td>\n",
       "      <td>0.183769</td>\n",
       "      <td>s_length</td>\n",
       "      <td>8.883252e-12</td>\n",
       "      <td>0.183769</td>\n",
       "      <td>[0]</td>\n",
       "      <td>[2.0]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>s_width</th>\n",
       "      <td>0.881502</td>\n",
       "      <td>0.204787</td>\n",
       "      <td>-0.260231</td>\n",
       "      <td>virginica</td>\n",
       "      <td>0.183769</td>\n",
       "      <td>s_width</td>\n",
       "      <td>3.249199e-03</td>\n",
       "      <td>0.884751</td>\n",
       "      <td>[1]</td>\n",
       "      <td>[3.2]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>p_length</th>\n",
       "      <td>0.999546</td>\n",
       "      <td>0.183399</td>\n",
       "      <td>-0.316458</td>\n",
       "      <td>virginica</td>\n",
       "      <td>0.183769</td>\n",
       "      <td>p_length</td>\n",
       "      <td>4.537308e-04</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>[2]</td>\n",
       "      <td>[1.8]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>p_width</th>\n",
       "      <td>0.462824</td>\n",
       "      <td>0.397061</td>\n",
       "      <td>-0.047643</td>\n",
       "      <td>virginica</td>\n",
       "      <td>0.183769</td>\n",
       "      <td>p_width</td>\n",
       "      <td>2.658730e-20</td>\n",
       "      <td>0.462824</td>\n",
       "      <td>[3]</td>\n",
       "      <td>[2.4]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                CI        CU     Cinfl    outname    outval   feature  \\\n",
       "s_length  0.183769  1.000000  0.091885  virginica  0.183769  s_length   \n",
       "s_width   0.881502  0.204787 -0.260231  virginica  0.183769   s_width   \n",
       "p_length  0.999546  0.183399 -0.316458  virginica  0.183769  p_length   \n",
       "p_width   0.462824  0.397061 -0.047643  virginica  0.183769   p_width   \n",
       "\n",
       "                  ymin      ymax inputs invals  neutralCU target_concept  \\\n",
       "s_length  8.883252e-12  0.183769    [0]  [2.0]        0.5           None   \n",
       "s_width   3.249199e-03  0.884751    [1]  [3.2]        0.5           None   \n",
       "p_length  4.537308e-04  1.000000    [2]  [1.8]        0.5           None   \n",
       "p_width   2.658730e-20  0.462824    [3]  [2.4]        0.5           None   \n",
       "\n",
       "         target_inputs  \n",
       "s_length          None  \n",
       "s_width           None  \n",
       "p_length          None  \n",
       "p_width           None  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "s_length   -0.183765\n",
       "s_width     0.520462\n",
       "p_length    0.632915\n",
       "p_width     0.419168\n",
       "dtype: float64"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "CIUres1 = CIU_iris.explain(iris_instance, output_inds=1)\n",
    "CIUres2 = CIU_iris.explain(iris_instance, output_inds=2)\n",
    "display(CIUres1)\n",
    "display(CIUres2)\n",
    "contrastive = contrastive_ciu(CIUres1, CIUres2)\n",
    "display(contrastive)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then the explanation:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ciuplots.plot_contrastive(CIUres1, CIUres2, main=\"Why is this Iris flower a Versicolor rather than a Virginica?\", figsize=(6,2))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The question can also be asked the other way around:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "CIUres2 = CIU_iris.explain(iris_instance, output_inds=0)\n",
    "ciuplots.plot_contrastive(CIUres2, CIUres1, main=\"Why is this not a Setosa but rather a Versicolor?\", figsize=(6,2))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Boston housing\n",
    "\n",
    "Start by loading the data and training a GB model on it. Also create CIU object and get CIU values for all the instances of interest. For the whole dataset this will typically take over a minute. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "<>:4: SyntaxWarning: invalid escape sequence '\\s'\n",
      "<>:4: SyntaxWarning: invalid escape sequence '\\s'\n",
      "/var/folders/65/868tj1315js3sr7vp6rnvvs00000gp/T/ipykernel_61516/3208799844.py:4: SyntaxWarning: invalid escape sequence '\\s'\n",
      "  raw_df = pd.read_csv(data_url, sep=\"\\s+\", skiprows=22, header=None)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "data_url = \"http://lib.stat.cmu.edu/datasets/boston\"\n",
    "raw_df = pd.read_csv(data_url, sep=\"\\s+\", skiprows=22, header=None)\n",
    "data = np.hstack([raw_df.values[::2, :], raw_df.values[1::2, :2]])\n",
    "boston_target = raw_df.values[1::2, 2]\n",
    "boston_data = pd.DataFrame(data)\n",
    "boston_data.columns = ['CRIM', 'ZN', 'INDUS', 'CHAS', 'NOX', 'RM', 'AGE', 'DIS', 'RAD', 'TAX', 'PTRATIO', 'B', 'LSTAT']\n",
    "np.random.seed(26) # We want to always get the same Random Forest model here.\n",
    "X_train, X_test, y_train, y_test = train_test_split(boston_data, boston_target, test_size=0.3, random_state=123)\n",
    "boston_xgb_model = xgb.XGBRegressor(objective ='reg:squarederror', colsample_bytree = 0.3, learning_rate = 0.1,max_depth = 5, alpha = 10, n_estimators = 10)\n",
    "boston_xgb_model.fit(X_train,y_train)\n",
    "out_minmaxs = pd.DataFrame({'mins': [min(y_train)], 'maxs': max(y_train)})\n",
    "out_minmaxs.index = ['Price']\n",
    "CIU_boston = ciu.CIU(boston_xgb_model.predict, ['Price'], data=X_train, out_minmaxs=out_minmaxs)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>CI</th>\n",
       "      <th>CU</th>\n",
       "      <th>Cinfl</th>\n",
       "      <th>outname</th>\n",
       "      <th>outval</th>\n",
       "      <th>feature</th>\n",
       "      <th>ymin</th>\n",
       "      <th>ymax</th>\n",
       "      <th>inputs</th>\n",
       "      <th>invals</th>\n",
       "      <th>neutralCU</th>\n",
       "      <th>target_concept</th>\n",
       "      <th>target_inputs</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>CRIM</th>\n",
       "      <td>0.007526</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.003763</td>\n",
       "      <td>Price</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>CRIM</td>\n",
       "      <td>31.792238</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>[0]</td>\n",
       "      <td>[0.03578]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ZN</th>\n",
       "      <td>0.017614</td>\n",
       "      <td>0.562698</td>\n",
       "      <td>0.001104</td>\n",
       "      <td>Price</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>ZN</td>\n",
       "      <td>31.684902</td>\n",
       "      <td>32.477524</td>\n",
       "      <td>[1]</td>\n",
       "      <td>[20.0]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>INDUS</th>\n",
       "      <td>0.046109</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.023055</td>\n",
       "      <td>Price</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>INDUS</td>\n",
       "      <td>30.055994</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>[2]</td>\n",
       "      <td>[3.33]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CHAS</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>Price</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>CHAS</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>[3]</td>\n",
       "      <td>[0.0]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NOX</th>\n",
       "      <td>0.025340</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.012670</td>\n",
       "      <td>Price</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>NOX</td>\n",
       "      <td>30.990591</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>[4]</td>\n",
       "      <td>[0.4429]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>RM</th>\n",
       "      <td>0.074985</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.037493</td>\n",
       "      <td>Price</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>RM</td>\n",
       "      <td>28.756565</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>[5]</td>\n",
       "      <td>[7.82]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AGE</th>\n",
       "      <td>0.013894</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.006947</td>\n",
       "      <td>Price</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>AGE</td>\n",
       "      <td>31.505692</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>[6]</td>\n",
       "      <td>[64.5]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>DIS</th>\n",
       "      <td>0.063518</td>\n",
       "      <td>0.922246</td>\n",
       "      <td>0.026820</td>\n",
       "      <td>Price</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>DIS</td>\n",
       "      <td>29.494846</td>\n",
       "      <td>32.353153</td>\n",
       "      <td>[7]</td>\n",
       "      <td>[4.6947]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>RAD</th>\n",
       "      <td>0.024754</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.012377</td>\n",
       "      <td>Price</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>RAD</td>\n",
       "      <td>31.016977</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>[8]</td>\n",
       "      <td>[5.0]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TAX</th>\n",
       "      <td>0.037765</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.018882</td>\n",
       "      <td>Price</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>TAX</td>\n",
       "      <td>30.431486</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>[9]</td>\n",
       "      <td>[216.0]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>PTRATIO</th>\n",
       "      <td>0.040333</td>\n",
       "      <td>0.396339</td>\n",
       "      <td>-0.004181</td>\n",
       "      <td>Price</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>PTRATIO</td>\n",
       "      <td>31.411566</td>\n",
       "      <td>33.226536</td>\n",
       "      <td>[10]</td>\n",
       "      <td>[14.9]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>B</th>\n",
       "      <td>0.023876</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.011938</td>\n",
       "      <td>Price</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>B</td>\n",
       "      <td>31.056467</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>[11]</td>\n",
       "      <td>[387.31]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>LSTAT</th>\n",
       "      <td>0.090572</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.045286</td>\n",
       "      <td>Price</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>LSTAT</td>\n",
       "      <td>28.055189</td>\n",
       "      <td>32.130909</td>\n",
       "      <td>[12]</td>\n",
       "      <td>[3.76]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               CI        CU     Cinfl outname     outval  feature       ymin  \\\n",
       "CRIM     0.007526  1.000000  0.003763   Price  32.130909     CRIM  31.792238   \n",
       "ZN       0.017614  0.562698  0.001104   Price  32.130909       ZN  31.684902   \n",
       "INDUS    0.046109  1.000000  0.023055   Price  32.130909    INDUS  30.055994   \n",
       "CHAS     0.000000  0.000000 -0.000000   Price  32.130909     CHAS  32.130909   \n",
       "NOX      0.025340  1.000000  0.012670   Price  32.130909      NOX  30.990591   \n",
       "RM       0.074985  1.000000  0.037493   Price  32.130909       RM  28.756565   \n",
       "AGE      0.013894  1.000000  0.006947   Price  32.130909      AGE  31.505692   \n",
       "DIS      0.063518  0.922246  0.026820   Price  32.130909      DIS  29.494846   \n",
       "RAD      0.024754  1.000000  0.012377   Price  32.130909      RAD  31.016977   \n",
       "TAX      0.037765  1.000000  0.018882   Price  32.130909      TAX  30.431486   \n",
       "PTRATIO  0.040333  0.396339 -0.004181   Price  32.130909  PTRATIO  31.411566   \n",
       "B        0.023876  1.000000  0.011938   Price  32.130909        B  31.056467   \n",
       "LSTAT    0.090572  1.000000  0.045286   Price  32.130909    LSTAT  28.055189   \n",
       "\n",
       "              ymax inputs     invals  neutralCU target_concept target_inputs  \n",
       "CRIM     32.130909    [0]  [0.03578]        0.5           None          None  \n",
       "ZN       32.477524    [1]     [20.0]        0.5           None          None  \n",
       "INDUS    32.130909    [2]     [3.33]        0.5           None          None  \n",
       "CHAS     32.130909    [3]      [0.0]        0.5           None          None  \n",
       "NOX      32.130909    [4]   [0.4429]        0.5           None          None  \n",
       "RM       32.130909    [5]     [7.82]        0.5           None          None  \n",
       "AGE      32.130909    [6]     [64.5]        0.5           None          None  \n",
       "DIS      32.353153    [7]   [4.6947]        0.5           None          None  \n",
       "RAD      32.130909    [8]      [5.0]        0.5           None          None  \n",
       "TAX      32.130909    [9]    [216.0]        0.5           None          None  \n",
       "PTRATIO  33.226536   [10]     [14.9]        0.5           None          None  \n",
       "B        32.130909   [11]   [387.31]        0.5           None          None  \n",
       "LSTAT    32.130909   [12]     [3.76]        0.5           None          None  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>CI</th>\n",
       "      <th>CU</th>\n",
       "      <th>Cinfl</th>\n",
       "      <th>outname</th>\n",
       "      <th>outval</th>\n",
       "      <th>feature</th>\n",
       "      <th>ymin</th>\n",
       "      <th>ymax</th>\n",
       "      <th>inputs</th>\n",
       "      <th>invals</th>\n",
       "      <th>neutralCU</th>\n",
       "      <th>target_concept</th>\n",
       "      <th>target_inputs</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>CRIM</th>\n",
       "      <td>0.016477</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-0.008239</td>\n",
       "      <td>Price</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>CRIM</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>29.542480</td>\n",
       "      <td>[0]</td>\n",
       "      <td>[0.0315]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ZN</th>\n",
       "      <td>0.017232</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.008616</td>\n",
       "      <td>Price</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>ZN</td>\n",
       "      <td>28.025560</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>[1]</td>\n",
       "      <td>[95.0]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>INDUS</th>\n",
       "      <td>0.026466</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.013233</td>\n",
       "      <td>Price</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>INDUS</td>\n",
       "      <td>27.610043</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>[2]</td>\n",
       "      <td>[1.47]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CHAS</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-0.000000</td>\n",
       "      <td>Price</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>CHAS</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>[3]</td>\n",
       "      <td>[0.0]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NOX</th>\n",
       "      <td>0.021159</td>\n",
       "      <td>0.790618</td>\n",
       "      <td>0.006149</td>\n",
       "      <td>Price</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>NOX</td>\n",
       "      <td>28.048223</td>\n",
       "      <td>29.000360</td>\n",
       "      <td>[4]</td>\n",
       "      <td>[0.403]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>RM</th>\n",
       "      <td>0.074985</td>\n",
       "      <td>0.322703</td>\n",
       "      <td>-0.013295</td>\n",
       "      <td>Price</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>RM</td>\n",
       "      <td>27.712090</td>\n",
       "      <td>31.086432</td>\n",
       "      <td>[5]</td>\n",
       "      <td>[6.975]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AGE</th>\n",
       "      <td>0.021635</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.010818</td>\n",
       "      <td>Price</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>AGE</td>\n",
       "      <td>27.827423</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>[6]</td>\n",
       "      <td>[15.3]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>DIS</th>\n",
       "      <td>0.050681</td>\n",
       "      <td>0.076359</td>\n",
       "      <td>-0.021470</td>\n",
       "      <td>Price</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>DIS</td>\n",
       "      <td>28.626854</td>\n",
       "      <td>30.907480</td>\n",
       "      <td>[7]</td>\n",
       "      <td>[7.6534]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>RAD</th>\n",
       "      <td>0.024887</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.012443</td>\n",
       "      <td>Price</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>RAD</td>\n",
       "      <td>27.681095</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>[8]</td>\n",
       "      <td>[3.0]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TAX</th>\n",
       "      <td>0.039942</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.019971</td>\n",
       "      <td>Price</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>TAX</td>\n",
       "      <td>27.003613</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>[9]</td>\n",
       "      <td>[402.0]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>PTRATIO</th>\n",
       "      <td>0.052906</td>\n",
       "      <td>0.792363</td>\n",
       "      <td>0.015468</td>\n",
       "      <td>Price</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>PTRATIO</td>\n",
       "      <td>26.914570</td>\n",
       "      <td>29.295336</td>\n",
       "      <td>[10]</td>\n",
       "      <td>[17.0]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>B</th>\n",
       "      <td>0.008612</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-0.004306</td>\n",
       "      <td>Price</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>B</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>29.188540</td>\n",
       "      <td>[11]</td>\n",
       "      <td>[396.9]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>LSTAT</th>\n",
       "      <td>0.126379</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.063189</td>\n",
       "      <td>Price</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>LSTAT</td>\n",
       "      <td>23.113947</td>\n",
       "      <td>28.801001</td>\n",
       "      <td>[12]</td>\n",
       "      <td>[4.56]</td>\n",
       "      <td>0.5</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               CI        CU     Cinfl outname     outval  feature       ymin  \\\n",
       "CRIM     0.016477  0.000000 -0.008239   Price  28.801001     CRIM  28.801001   \n",
       "ZN       0.017232  1.000000  0.008616   Price  28.801001       ZN  28.025560   \n",
       "INDUS    0.026466  1.000000  0.013233   Price  28.801001    INDUS  27.610043   \n",
       "CHAS     0.000000  0.000000 -0.000000   Price  28.801001     CHAS  28.801001   \n",
       "NOX      0.021159  0.790618  0.006149   Price  28.801001      NOX  28.048223   \n",
       "RM       0.074985  0.322703 -0.013295   Price  28.801001       RM  27.712090   \n",
       "AGE      0.021635  1.000000  0.010818   Price  28.801001      AGE  27.827423   \n",
       "DIS      0.050681  0.076359 -0.021470   Price  28.801001      DIS  28.626854   \n",
       "RAD      0.024887  1.000000  0.012443   Price  28.801001      RAD  27.681095   \n",
       "TAX      0.039942  1.000000  0.019971   Price  28.801001      TAX  27.003613   \n",
       "PTRATIO  0.052906  0.792363  0.015468   Price  28.801001  PTRATIO  26.914570   \n",
       "B        0.008612  0.000000 -0.004306   Price  28.801001        B  28.801001   \n",
       "LSTAT    0.126379  1.000000  0.063189   Price  28.801001    LSTAT  23.113947   \n",
       "\n",
       "              ymax inputs    invals  neutralCU target_concept target_inputs  \n",
       "CRIM     29.542480    [0]  [0.0315]        0.5           None          None  \n",
       "ZN       28.801001    [1]    [95.0]        0.5           None          None  \n",
       "INDUS    28.801001    [2]    [1.47]        0.5           None          None  \n",
       "CHAS     28.801001    [3]     [0.0]        0.5           None          None  \n",
       "NOX      29.000360    [4]   [0.403]        0.5           None          None  \n",
       "RM       31.086432    [5]   [6.975]        0.5           None          None  \n",
       "AGE      28.801001    [6]    [15.3]        0.5           None          None  \n",
       "DIS      30.907480    [7]  [7.6534]        0.5           None          None  \n",
       "RAD      28.801001    [8]     [3.0]        0.5           None          None  \n",
       "TAX      28.801001    [9]   [402.0]        0.5           None          None  \n",
       "PTRATIO  29.295336   [10]    [17.0]        0.5           None          None  \n",
       "B        29.188540   [11]   [396.9]        0.5           None          None  \n",
       "LSTAT    28.801001   [12]    [4.56]        0.5           None          None  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "CRIM       0.007526\n",
       "ZN        -0.007703\n",
       "INDUS      0.000000\n",
       "CHAS       0.000000\n",
       "NOX        0.005306\n",
       "RM         0.050787\n",
       "AGE        0.000000\n",
       "DIS        0.053729\n",
       "RAD        0.000000\n",
       "TAX        0.000000\n",
       "PTRATIO   -0.015973\n",
       "B          0.023876\n",
       "LSTAT      0.000000\n",
       "dtype: float64"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "inst1 = X_test.iloc[[2]]\n",
    "inst2 = X_test.iloc[[4]]\n",
    "CIUres1 = CIU_boston.explain(inst1)\n",
    "CIUres2 = CIU_boston.explain(inst2)\n",
    "display(CIUres1)\n",
    "display(CIUres2)\n",
    "contrastive = contrastive_ciu(CIUres1, CIUres2)\n",
    "display(contrastive)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ciuplots.plot_contrastive(CIUres1, CIUres2, main=\"Why is instance #2 more expensive than instance #4?\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Ames housing\n",
    "\n",
    "We next produce a contrastive explanation for why Ames instance #433 ($477919, previous Figures) is predicted to be more expensive than instance #1638 ($439581). Contrastive values are in the range [−1, 1] by definition, so the differences between these two Ames instances are rather small. The two instances have been selected to be quite similar because contrastive questions/explanations are often asked as \"why did you buy that car, isn't that one just as good or better?\". "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "py-ciu",
   "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": "3.12.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}