couples-lime.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "from sklearn.pipeline import make_pipeline, Pipeline\n",
    "from sklearn.preprocessing import StandardScaler, OneHotEncoder, LabelEncoder\n",
    "from sklearn.pipeline import FeatureUnion\n",
    "from sklearn.base import BaseEstimator, TransformerMixin\n",
    "from sklearn.model_selection import train_test_split, GridSearchCV\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.metrics import f1_score\n",
    "\n",
    "from lime.lime_tabular import LimeTabularExplainer\n",
    "\n",
    "pd.set_option('display.max_columns', None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Class to help select categorical vs. continuous data as part of the pipeline (see below)\n",
    "class DataSelector(BaseEstimator, TransformerMixin):\n",
    "    '''Select columns of numpy arrays based on attribute_indices.'''\n",
    "\n",
    "    def __init__(self, attribute_indices):\n",
    "        self.attribute_indices = attribute_indices\n",
    "\n",
    "    def fit(self, X, y=None):\n",
    "        return self\n",
    "\n",
    "    def transform(self, X):\n",
    "        return np.array(X)[:,self.attribute_indices]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Load (preprocessed) data\n",
    "# \n",
    "# The raw data was downloaded from https://data.stanford.edu/hcmst and preprocessed.\n",
    "# We combined data sets collected across several years, we transformed select variables \n",
    "# (e.g., partner_education to be at the same level of granularity as education),\n",
    "# and added variables like the absolute age difference, education difference, etc.\n",
    "# Finally, we determined whether couples were still together (i.e., our labels).\n",
    "#\n",
    "# We provide the preprocessed data as a csv file in the same repo as this notebook.\n",
    "\n",
    "df = pd.read_csv('couples.csv')\n",
    "\n",
    "# Order features (numeric first, categorical second) since it's convenient later\n",
    "feature_order = ['age',\n",
    "                 'partner_age',\n",
    "                 'age_diff_abs',\n",
    "                 'children',\n",
    "                 'visits_relatives',\n",
    "                 'education',\n",
    "                 'marital_status',\n",
    "                 'partner_education',\n",
    "                 'gender',\n",
    "                 'house',\n",
    "                 'income',\n",
    "                 'msa',\n",
    "                 'rent',\n",
    "                 'political',\n",
    "                 'religion',\n",
    "                 'work',\n",
    "                 'gender_older',\n",
    "                 'education_difference',\n",
    "                 'success']\n",
    "\n",
    "data = df[feature_order]\n",
    "data = data[data.house != 'boat, rv, van, etc.'] # only one data point with this value, discard\n",
    "\n",
    "labels = data.pop('success')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>age</th>\n",
       "      <th>partner_age</th>\n",
       "      <th>age_diff_abs</th>\n",
       "      <th>children</th>\n",
       "      <th>visits_relatives</th>\n",
       "      <th>education</th>\n",
       "      <th>marital_status</th>\n",
       "      <th>partner_education</th>\n",
       "      <th>gender</th>\n",
       "      <th>house</th>\n",
       "      <th>income</th>\n",
       "      <th>msa</th>\n",
       "      <th>rent</th>\n",
       "      <th>political</th>\n",
       "      <th>religion</th>\n",
       "      <th>work</th>\n",
       "      <th>gender_older</th>\n",
       "      <th>education_difference</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>52</td>\n",
       "      <td>48</td>\n",
       "      <td>4</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>bachelor's degree or higher</td>\n",
       "      <td>living with partner</td>\n",
       "      <td>some college</td>\n",
       "      <td>female</td>\n",
       "      <td>a building with 2 or more apartments</td>\n",
       "      <td>$20,000 to $24,999</td>\n",
       "      <td>metro</td>\n",
       "      <td>rented for cash</td>\n",
       "      <td>democrat</td>\n",
       "      <td>catholic</td>\n",
       "      <td>working - as a paid employee</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>28</td>\n",
       "      <td>30</td>\n",
       "      <td>2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>bachelor's degree or higher</td>\n",
       "      <td>living with partner</td>\n",
       "      <td>bachelor's degree or higher</td>\n",
       "      <td>female</td>\n",
       "      <td>a building with 2 or more apartments</td>\n",
       "      <td>$40,000 to $49,999</td>\n",
       "      <td>metro</td>\n",
       "      <td>rented for cash</td>\n",
       "      <td>democrat</td>\n",
       "      <td>jewish</td>\n",
       "      <td>working - as a paid employee</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>31</td>\n",
       "      <td>40</td>\n",
       "      <td>9</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>some college</td>\n",
       "      <td>never married</td>\n",
       "      <td>high school</td>\n",
       "      <td>male</td>\n",
       "      <td>a building with 2 or more apartments</td>\n",
       "      <td>$40,000 to $49,999</td>\n",
       "      <td>metro</td>\n",
       "      <td>owned or being bought by you or someone in you...</td>\n",
       "      <td>democrat</td>\n",
       "      <td>other non-christian, please specify:</td>\n",
       "      <td>working - as a paid employee</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>53</td>\n",
       "      <td>55</td>\n",
       "      <td>2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>bachelor's degree or higher</td>\n",
       "      <td>living with partner</td>\n",
       "      <td>bachelor's degree or higher</td>\n",
       "      <td>male</td>\n",
       "      <td>a one-family house detached from any other house</td>\n",
       "      <td>$125,000 to $149,999</td>\n",
       "      <td>metro</td>\n",
       "      <td>owned or being bought by you or someone in you...</td>\n",
       "      <td>democrat</td>\n",
       "      <td>protestant (e.g., methodist, lutheran, presbyt...</td>\n",
       "      <td>working - as a paid employee</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>58</td>\n",
       "      <td>51</td>\n",
       "      <td>7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>bachelor's degree or higher</td>\n",
       "      <td>separated</td>\n",
       "      <td>bachelor's degree or higher</td>\n",
       "      <td>male</td>\n",
       "      <td>a building with 2 or more apartments</td>\n",
       "      <td>$15,000 to $19,999</td>\n",
       "      <td>metro</td>\n",
       "      <td>rented for cash</td>\n",
       "      <td>democrat</td>\n",
       "      <td>protestant (e.g., methodist, lutheran, presbyt...</td>\n",
       "      <td>working - as a paid employee</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   age  partner_age  age_diff_abs  children  visits_relatives  \\\n",
       "0   52           48             4       0.0                 0   \n",
       "1   28           30             2       0.0                 0   \n",
       "2   31           40             9       0.0                 1   \n",
       "3   53           55             2       0.0                 1   \n",
       "4   58           51             7       0.0                 0   \n",
       "\n",
       "                     education       marital_status  \\\n",
       "0  bachelor's degree or higher  living with partner   \n",
       "1  bachelor's degree or higher  living with partner   \n",
       "2                 some college        never married   \n",
       "3  bachelor's degree or higher  living with partner   \n",
       "4  bachelor's degree or higher            separated   \n",
       "\n",
       "             partner_education  gender  \\\n",
       "0                 some college  female   \n",
       "1  bachelor's degree or higher  female   \n",
       "2                  high school    male   \n",
       "3  bachelor's degree or higher    male   \n",
       "4  bachelor's degree or higher    male   \n",
       "\n",
       "                                              house                income  \\\n",
       "0              a building with 2 or more apartments    $20,000 to $24,999   \n",
       "1              a building with 2 or more apartments    $40,000 to $49,999   \n",
       "2              a building with 2 or more apartments    $40,000 to $49,999   \n",
       "3  a one-family house detached from any other house  $125,000 to $149,999   \n",
       "4              a building with 2 or more apartments    $15,000 to $19,999   \n",
       "\n",
       "     msa                                               rent political  \\\n",
       "0  metro                                    rented for cash  democrat   \n",
       "1  metro                                    rented for cash  democrat   \n",
       "2  metro  owned or being bought by you or someone in you...  democrat   \n",
       "3  metro  owned or being bought by you or someone in you...  democrat   \n",
       "4  metro                                    rented for cash  democrat   \n",
       "\n",
       "                                            religion  \\\n",
       "0                                           catholic   \n",
       "1                                             jewish   \n",
       "2               other non-christian, please specify:   \n",
       "3  protestant (e.g., methodist, lutheran, presbyt...   \n",
       "4  protestant (e.g., methodist, lutheran, presbyt...   \n",
       "\n",
       "                           work  gender_older  education_difference  \n",
       "0  working - as a paid employee             1                     1  \n",
       "1  working - as a paid employee             0                     0  \n",
       "2  working - as a paid employee             1                     1  \n",
       "3  working - as a paid employee             1                     0  \n",
       "4  working - as a paid employee             0                     0  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Take a peak at the data\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Define categorical names and indices\n",
    "categorical_features = list(data.columns[5:])\n",
    "categorical_idx = list(range(5, len(data.columns)))\n",
    "continuous_features = list(data.columns[0:5])\n",
    "continuous_idx = list(range(0,5))\n",
    "\n",
    "X = data.values\n",
    "\n",
    "# Get feature names and their values for categorical data (needed for LIME)\n",
    "categorical_names = {}\n",
    "for idx, feature in zip(categorical_idx, categorical_features):\n",
    "    le = LabelEncoder()\n",
    "    X[:, idx] = le.fit_transform(X[:, idx])\n",
    "    categorical_names[idx] = le.classes_\n",
    "\n",
    "# To suppress a warning later (not strictly necessary)\n",
    "X = X.astype(float)\n",
    "\n",
    "# Train test split\n",
    "train, test, labels_train, labels_test = train_test_split(\n",
    "    X, labels, train_size=0.70, random_state=42\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Preprocessing pipeline\n",
    "#      \n",
    "# LIME needs a function that takes raw inputs and returns a prediction (see below).     \n",
    "# We use sklearn's pipeline to handle preprocessing, it simplifies the interaction with LIME (see below). \n",
    "# There are several ways to build this pipeline. For demo purposes, we here show the verbose option (and we\n",
    "# avoid scaling one-hot encoded features).\n",
    "\n",
    "continuous_pipeline = Pipeline([\n",
    "    ('selector', DataSelector(continuous_idx)),\n",
    "    ('scaler', StandardScaler()),\n",
    "    ])\n",
    "\n",
    "categorical_pipeline = Pipeline([\n",
    "    ('selector', DataSelector(categorical_idx)),\n",
    "    ('encoder', OneHotEncoder(sparse=False)),\n",
    "    ])\n",
    "\n",
    "preprocessing_pipeline = FeatureUnion(transformer_list=[\n",
    "    (\"continuous_pipeline\", continuous_pipeline),\n",
    "    (\"categorical_pipeline\", categorical_pipeline),\n",
    "    ])\n",
    "\n",
    "# There are less verbose alternatives, especially if we scale one-hot encoded features,\n",
    "# an accepted practice in the machine learning community:\n",
    "#\n",
    "#     preprocessing_pipeline = Pipeline([\n",
    "#        ('onehotencoder', OneHotEncoder(categorical_features=categorical_idx, sparse=False)),\n",
    "#        ('scaler', StandardScaler())\n",
    "#     ])\n",
    "#\n",
    "# Finally, instead of the low-level Pipeline constructor, we can use sklearn's makepipeline:\n",
    "#\n",
    "#     preprocessing_pipeline = make_pipeline(\n",
    "#         OneHotEncoder(categorical_features=categorical_idx, sparse=False),\n",
    "#         StandardScaler()\n",
    "#     )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Pipeline(steps=[('featureunion', FeatureUnion(n_jobs=1,\n",
       "       transformer_list=[('continuous_pipeline', Pipeline(steps=[('selector', DataSelector(attribute_indices=[0, 1, 2, 3, 4])), ('scaler', StandardScaler(copy=True, with_mean=True, with_std=True))])), ('categorical_pipeline', Pipeline(steps=[('selector'...*n_jobs', refit=True, return_train_score=True,\n",
       "       scoring='neg_mean_squared_error', verbose=0))])"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Set up the model and GridSearch for random forest hyperparameter tuning\n",
    "param_grid = [{'n_estimators': [16, 20, 24],\n",
    "               'max_features': [4, 8, 12, 16],\n",
    "               'min_samples_split': [8, 12, 16],\n",
    "               'max_depth': [15, 20, 25]}]\n",
    "\n",
    "rf = RandomForestClassifier(class_weight='balanced')\n",
    "grid_search = GridSearchCV(rf, param_grid, cv=5, scoring='neg_mean_squared_error')\n",
    "\n",
    "# Extend the preprocessing pipeline to include random forest and grid search\n",
    "pipeline = make_pipeline(preprocessing_pipeline, grid_search)\n",
    "\n",
    "# Fit the model and tune the hyperparameters\n",
    "pipeline.fit(train, labels_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'max_depth': 20, 'max_features': 16, 'min_samples_split': 16, 'n_estimators': 24}\n"
     ]
    }
   ],
   "source": [
    "# Hyperparameters found by GridSearchCV\n",
    "print(pipeline.named_steps['gridsearchcv'].best_params_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "F1 on train: 0.928947368421\n",
      "F1 on test: 0.885885885886\n"
     ]
    }
   ],
   "source": [
    "# Evalute random forest classifier on training data (it overfits, small sample size)\n",
    "y_predict = pipeline.predict(train)\n",
    "f1 = f1_score(labels_train, y_predict)\n",
    "print('F1 on train:', f1)\n",
    "\n",
    "# Evalute random forest classifier on test data\n",
    "y_predict = pipeline.predict(test)\n",
    "f1 = f1_score(labels_test, y_predict)\n",
    "print('F1 on test:', f1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Feature ranking:\n",
      " 1. feature age (0.225348)\n",
      " 2. feature partner_age (0.116904)\n",
      " 3. feature age_diff_abs (0.087704)\n",
      " 4. feature children (0.060545)\n",
      " 5. feature visits_relatives (0.058321)\n",
      " 6. feature education (0.054985)\n",
      " 7. feature marital_status (0.034843)\n",
      " 8. feature partner_education (0.022823)\n",
      " 9. feature gender (0.020148)\n",
      "10. feature house (0.015524)\n",
      "11. feature income (0.013082)\n",
      "12. feature msa (0.011647)\n",
      "13. feature rent (0.011330)\n",
      "14. feature political (0.010574)\n",
      "15. feature religion (0.009844)\n",
      "16. feature work (0.009569)\n",
      "17. feature gender_older (0.009383)\n",
      "18. feature education_difference (0.009010)\n"
     ]
    }
   ],
   "source": [
    "# Get feature importances of random forest model (\"global interpretability\")\n",
    "best_estimator = pipeline.named_steps['gridsearchcv'].best_estimator_\n",
    "\n",
    "importances = best_estimator.feature_importances_\n",
    "std = np.std([tree.feature_importances_ for tree in best_estimator.estimators_],\n",
    "             axis=0)\n",
    "indices = np.argsort(importances)[::-1]\n",
    "\n",
    "print(\"Feature ranking:\")\n",
    "feature_names = data.columns\n",
    "for f in range(data.shape[1]):\n",
    "    print(\"%2d. feature %s (%f)\" %\n",
    "          (f + 1, feature_names[f], importances[indices[f]]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Use LIME to explain individual predictions, initialize explainer object\n",
    "explainer = LimeTabularExplainer(\n",
    "    train,\n",
    "    class_names=['BrokeUp', 'StayedTogether'],\n",
    "    feature_names=list(data.columns),\n",
    "    categorical_features=categorical_idx,\n",
    "    categorical_names=categorical_names,\n",
    "    discretize_continuous=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Couples probability of staying together: 0.927837385823\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAArcAAAEICAYAAABF6dxKAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n5FaSJEm98f8DKbzFKpuC8cQAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x111667be0>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Explain a prediction (\"local interpretability\"): \n",
    "# Now we see that the pipeline that takes raw data and returns the prediction \n",
    "# of the trained model now comes in conveniently.\n",
    "example = 3\n",
    "exp = explainer.explain_instance(test[example], pipeline.predict_proba, num_features=5)\n",
    "print('Couples probability of staying together:', exp.predict_proba[1])\n",
    "exp.as_pyplot_figure()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Couples probability of staying together: 0.692087931046\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAArcAAAEICAYAAABF6dxKAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1184b0358>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Explain another prediction (\"local interpretability\"): \n",
    "example = 13\n",
    "exp = explainer.explain_instance(test[example], pipeline.predict_proba, num_features=5)\n",
    "print('Couples probability of staying together:', exp.predict_proba[1])\n",
    "exp.as_pyplot_figure()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# and we see differences in explaining the model's predictions."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Couples probability of staying together: 0.73871988093\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAlYAAAEICAYAAACdyboFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nJ1ZmZmZmJfn/CkgxaJb73MAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x118504828>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Using LIME for for relationship management (not advised): Current chance of relationship success.\n",
    "current = [34, 36, 2, 0, 1, 0, 1, 0, 0, 0, 18, 0, 2, 0, 8, 4, 0, 0]\n",
    "exp = explainer.explain_instance(np.array(current), pipeline.predict_proba, num_features=5)\n",
    "print('Couples probability of staying together:', exp.predict_proba[1])\n",
    "exp.as_pyplot_figure()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Couples probability of staying together: 0.73871988093\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAlYAAAEICAYAAACdyboFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11112e1d0>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Should I ask for a pay increase? It doesn't matter much.\n",
    "increase_income = [34, 36, 2, 0, 1, 0, 1, 0, 0, 0, 6, 0, 2, 0, 8, 4, 0, 0]\n",
    "exp = explainer.explain_instance(np.array(increase_income), pipeline.predict_proba, num_features=5)\n",
    "print('Couples probability of staying together:', exp.predict_proba[1])\n",
    "exp.as_pyplot_figure()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Couples probability of staying together: 0.767889240003\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAlYAAAEICAYAAACdyboFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x111c590b8>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Should I buy a house? Maybe?\n",
    "buy_house = [34, 36, 2, 0, 1, 0, 1, 0, 0, 0, 6, 0, 1, 0, 8, 4, 0, 0]\n",
    "exp = explainer.explain_instance(np.array(buy_house), pipeline.predict_proba, num_features=5)\n",
    "print('Couples probability of staying together:', exp.predict_proba[1])\n",
    "exp.as_pyplot_figure()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Couples probability of staying together: 0.982588631027\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAArcAAAEICAYAAABF6dxKAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x111c595c0>"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Really, it's best to get married.\n",
    "get_married = [34, 36, 2, 0, 1, 0, 2, 0, 0, 0, 6, 0, 1, 0, 8, 4, 0, 0]\n",
    "exp = explainer.explain_instance(np.array(get_married), pipeline.predict_proba, num_features=5)\n",
    "print('Couples probability of staying together:', exp.predict_proba[1])\n",
    "exp.as_pyplot_figure()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}