examples/Lime.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7cab43eb",
   "metadata": {},
   "outputs": [],
   "source": [
    "%pip install -r ../requirements-examples.txt --quiet"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "12f019c1",
   "metadata": {},
   "source": [
    "# LIME explanations"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5b6bb653",
   "metadata": {},
   "source": [
    "## Simple example\n",
    "\n",
    "We start by defining our black-box model, typically represented by\n",
    "\n",
    "$$\n",
    "f(\\mathbf{x}) = \\mathbf{y}\n",
    "$$\n",
    "\n",
    "Where $\\mathbf{x}=\\{x_1, x_2, \\dots,x_m\\}$ and $\\mathbf{y}=\\{y_1, y_2, \\dots,y_n\\}$.\n",
    "\n",
    "Our example toy model, in this case, takes an all-numerical input $\\mathbf{x}$ and return a $\\mathbf{y}$ of either `true` or `false` if the sum of the $\\mathbf{x}$ components is within a threshold $\\epsilon$ of a point $\\mathbf{C}$, that is:\n",
    "\n",
    "$$\n",
    "f(\\mathbf{x}, \\epsilon, \\mathbf{C})=\\begin{cases}\n",
    "\\text{true},\\qquad \\text{if}\\ \\mathbf{C}-\\epsilon<\\sum_{i=1}^m x_i <\\mathbf{C}+\\epsilon \\\\\n",
    "\\text{false},\\qquad \\text{otherwise}\n",
    "\\end{cases}\n",
    "$$\n",
    "\n",
    "This model is provided in the `TestUtils` module. We instantiate with a $\\mathbf{C}=500$ and $\\epsilon=1.0$."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "e20342d0",
   "metadata": {},
   "outputs": [],
   "source": [
    "from trustyai.utils import TestModels\n",
    "\n",
    "center = 10.0\n",
    "epsilon = 2.0\n",
    "\n",
    "model = TestModels.getSumThresholdModel(center, epsilon)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3ea83b33",
   "metadata": {},
   "source": [
    "Next we need to define a **goal**.\n",
    "If our model is $f(\\mathbf{x'})=\\mathbf{y'}$ we are then defining our $\\mathbf{y'}$ and the counterfactual result will be the $\\mathbf{x'}$ which satisfies $f(\\mathbf{x'})=\\mathbf{y'}$.\n",
    "\n",
    "We will define our goal as `true`, that is, the sum is withing the vicinity of a (to be defined) point $\\mathbf{C}$. The goal is a list of `Output` which take the following parameters\n",
    "\n",
    "- The feature name\n",
    "- The feature type\n",
    "- The feature value (wrapped in `Value`)\n",
    "- A confidence threshold, which we will leave at zero (no threshold)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "2b011c6a",
   "metadata": {},
   "outputs": [],
   "source": [
    "from trustyai.model import output\n",
    "\n",
    "decision = \"inside\"\n",
    "goal = [output(name=decision, dtype=\"bool\", value=True, score=0.0)]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5e38a38c",
   "metadata": {},
   "source": [
    "We will now define our initial features, $\\mathbf{x}$. Each feature can be instantiated by using `FeatureFactory` and in this case we want to use numerical features, so we'll use `FeatureFactory.newNumericalFeature`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "39946517",
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "from trustyai.model import feature\n",
    "\n",
    "features = [feature(name=f\"x{i+1}\", dtype=\"number\", value=random.random()*10.0) for i in range(3)]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cd098cff",
   "metadata": {},
   "source": [
    "As we can see, the sum of of the features will not be within $\\epsilon$ (1.0) of $\\mathbf{C}$ (500.0). As such the model prediction will be `false`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "350595e3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Feature x1 has value 4.214275243409668\n",
      "Feature x2 has value 6.029859645751495\n",
      "Feature x3 has value 1.13146410299228\n",
      "\n",
      "Features sum is 11.375598992153444\n"
     ]
    }
   ],
   "source": [
    "feature_sum = 0.0\n",
    "for f in features:\n",
    "    value = f.value.as_number()\n",
    "    print(f\"Feature {f.name} has value {value}\")\n",
    "    feature_sum += value\n",
    "print(f\"\\nFeatures sum is {feature_sum}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "abf7a99c",
   "metadata": {},
   "source": [
    "We execute the model on the generated input and collect the output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "a8d3679a",
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Output{value=true, type=boolean, score=0.624401007846556, name='inside'}'"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from org.kie.trustyai.explainability.model import PredictionInput, PredictionOutput\n",
    "\n",
    "goals = model.predictAsync([PredictionInput(features)]).get()\n",
    "goals[0].outputs.get(0).toString()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "18cba3f3",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "We can now instantiate the **explainer** itself.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "158f797f",
   "metadata": {},
   "outputs": [],
   "source": [
    "from trustyai.explainers import LimeExplainer\n",
    "\n",
    "explainer = LimeExplainer(samples=10)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d7865286",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "We generate the **explanation** as a _dict : decision --> saliency_.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "bc4d2a55",
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "explanation = explainer.explain(inputs=features, outputs=goals[0].outputs, model=model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "de62e6d1",
   "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>Feature</th>\n",
       "      <th>Value</th>\n",
       "      <th>Saliency</th>\n",
       "      <th>Confidence</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>x1</td>\n",
       "      <td>4.214275</td>\n",
       "      <td>0.422487</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>x2</td>\n",
       "      <td>6.029860</td>\n",
       "      <td>0.474942</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>x3</td>\n",
       "      <td>1.131464</td>\n",
       "      <td>0.375759</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Feature     Value  Saliency  Confidence\n",
       "0      x1  4.214275  0.422487         0.0\n",
       "1      x2  6.029860  0.474942         0.0\n",
       "2      x3  1.131464  0.375759         0.0"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "explanation.as_dataframe()['inside']"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3ed76c1c",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "We inspect the saliency scores assigned by LIME to each feature"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e9b5d082",
   "metadata": {},
   "source": [
    "We generate the saliency graph with the builtin method `plot(decision)`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "d789aad1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1600x900 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "explanation.plot(decision)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1e4f109c",
   "metadata": {},
   "source": [
    "## Using Python models\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d5abc216",
   "metadata": {},
   "source": [
    "We will now show how to use a custom Python model with TrustyAI LIME implementation.\n",
    "\n",
    "The model will be an XGBoost one trained with the `credit-bias` dataset.\n",
    "\n",
    "For convenience, the model is pre-trained and serialised with `joblib` so that for this example we simply need to deserialised it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "17afda54",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/rui/.virtualenvs/trustyai-explainability-python-examples/lib/python3.11/site-packages/xgboost/compat.py:31: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.\n",
      "  from pandas import MultiIndex, Int64Index\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "XGBClassifier(base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
      "              colsample_bynode=1, colsample_bytree=1, gamma=0, gpu_id=-1,\n",
      "              importance_type='gain', interaction_constraints='',\n",
      "              learning_rate=0.07, max_delta_step=0, max_depth=8,\n",
      "              min_child_weight=1, missing=nan, monotone_constraints='()',\n",
      "              n_estimators=200, n_jobs=12, num_parallel_tree=1, random_state=27,\n",
      "              reg_alpha=0, reg_lambda=1, scale_pos_weight=0.9861206227457426,\n",
      "              seed=27, subsample=1, tree_method='exact', validate_parameters=1,\n",
      "              verbosity=None)\n"
     ]
    }
   ],
   "source": [
    "import joblib\n",
    "\n",
    "xg_model = joblib.load(\"models/credit-bias-xgboost.joblib\")\n",
    "print(xg_model)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4205418c",
   "metadata": {},
   "source": [
    "This model has as a single **output** a boolean `PaidLoan`, which will predict whether a certain loan applicant will repay the loan in time or not. The model is slightly more complex than the previous examples, with **input** features:\n",
    "\n",
    "|Input feature         | Type    | Note        |\n",
    "|----------------------|---------|-------------|\n",
    "|`NewCreditCustomer`   |boolean  ||\n",
    "|`Amount`              |numerical||\n",
    "|`Interest`            |numerical||\n",
    "|`LoanDuration`        |numerical|In months|\n",
    "|`Education`           |numerical|Level (1, 2, 3..)|\n",
    "|`NrOfDependants`      |numerical|Integer|\n",
    "|`EmploymentDurationCurrentEmployer`|numerical|Integer (years)|\n",
    "|`IncomeFromPrincipalEmployer`|numerical||\n",
    "|`IncomeFromPension`   |numerical||\n",
    "|`IncomeFromFamilyAllowance`|numerical||\n",
    "|`IncomeFromSocialWelfare`|numerical||\n",
    "|`IncomeFromLeavePay`|numerical||\n",
    "|`IncomeFromChildSupport`|numerical||\n",
    "|`IncomeOther`|numerical||\n",
    "|`ExistingLiabilities`|numerical|integer|\n",
    "|`RefinanceLiabilities`|numerical|integer|\n",
    "|`DebtToIncome`|numerical||\n",
    "|`FreeCash`|numerical||\n",
    "|`CreditScoreEeMini`|numerical|integer|\n",
    "|`NoOfPreviousLoansBeforeLoan`|numerical|integer|\n",
    "|`AmountOfPreviousLoansBeforeLoan`|numerical||\n",
    "|`PreviousRepaymentsBeforeLoan`|numerical||\n",
    "|`PreviousEarlyRepaymentsBefoleLoan`|numerical||\n",
    "|`PreviousEarlyRepaymentsCountBeforeLoan`|numerical|integer|\n",
    "|`Council_house`|boolean||\n",
    "|`Homeless`|boolean||\n",
    "|`Joint_ownership`|boolean||\n",
    "|`Joint_tenant`|boolean||\n",
    "|`Living_with_parents`|boolean||\n",
    "|`Mortgage`|boolean||\n",
    "|`Other`|boolean||\n",
    "|`Owner`|boolean||\n",
    "|`Owner_with_encumbrance`|boolean||\n",
    "|`Tenant`|boolean||\n",
    "|`Entrepreneur`|boolean||\n",
    "|`Fully`|boolean||\n",
    "|`Partially`|boolean||\n",
    "|`Retiree`|boolean||\n",
    "|`Self_employed`|boolean||\n",
    "\n",
    "Let's create and test a sample datapoint:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "48d05ff5",
   "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>NewCreditCustomer</th>\n",
       "      <th>Amount</th>\n",
       "      <th>Interest</th>\n",
       "      <th>LoanDuration</th>\n",
       "      <th>Education</th>\n",
       "      <th>NrOfDependants</th>\n",
       "      <th>EmploymentDurationCurrentEmployer</th>\n",
       "      <th>IncomeFromPrincipalEmployer</th>\n",
       "      <th>IncomeFromPension</th>\n",
       "      <th>IncomeFromFamilyAllowance</th>\n",
       "      <th>...</th>\n",
       "      <th>Mortgage</th>\n",
       "      <th>Other</th>\n",
       "      <th>Owner</th>\n",
       "      <th>Owner_with_encumbrance</th>\n",
       "      <th>Tenant</th>\n",
       "      <th>Entrepreneur</th>\n",
       "      <th>Fully</th>\n",
       "      <th>Partially</th>\n",
       "      <th>Retiree</th>\n",
       "      <th>Self_employed</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>False</td>\n",
       "      <td>2125.0</td>\n",
       "      <td>20.97</td>\n",
       "      <td>60.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>301.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1 rows × 39 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   NewCreditCustomer  Amount  Interest  LoanDuration  Education  \\\n",
       "0              False  2125.0     20.97          60.0        4.0   \n",
       "\n",
       "   NrOfDependants  EmploymentDurationCurrentEmployer  \\\n",
       "0             0.0                                6.0   \n",
       "\n",
       "   IncomeFromPrincipalEmployer  IncomeFromPension  IncomeFromFamilyAllowance  \\\n",
       "0                          0.0              301.0                        0.0   \n",
       "\n",
       "   ...  Mortgage  Other  Owner  Owner_with_encumbrance  Tenant  Entrepreneur  \\\n",
       "0  ...     False  False  False                    True    True         False   \n",
       "\n",
       "   Fully  Partially  Retiree  Self_employed  \n",
       "0  False      False     True          False  \n",
       "\n",
       "[1 rows x 39 columns]"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "feature_values = {\n",
    "        \"NewCreditCustomer\": False,\n",
    "        \"Amount\": 2125.0,\n",
    "        \"Interest\": 20.97,\n",
    "        \"LoanDuration\": 60.0,\n",
    "        \"Education\": 4.0,\n",
    "        \"NrOfDependants\": 0.0,\n",
    "        \"EmploymentDurationCurrentEmployer\": 6.0,\n",
    "        \"IncomeFromPrincipalEmployer\": 0.0,\n",
    "        \"IncomeFromPension\": 301.0,\n",
    "        \"IncomeFromFamilyAllowance\": 0.0,\n",
    "        \"IncomeFromSocialWelfare\": 53.0,\n",
    "        \"IncomeFromLeavePay\": 0.0,\n",
    "        \"IncomeFromChildSupport\": 0.0,\n",
    "        \"IncomeOther\": 0.0,\n",
    "        \"ExistingLiabilities\": 8.0,\n",
    "        \"RefinanceLiabilities\": 6.0,\n",
    "        \"DebtToIncome\": 26.29,\n",
    "        \"FreeCash\": 10.92,\n",
    "        \"CreditScoreEeMini\": 1000.0,\n",
    "        \"NoOfPreviousLoansBeforeLoan\": 1.0,\n",
    "        \"AmountOfPreviousLoansBeforeLoan\": 500.0,\n",
    "        \"PreviousRepaymentsBeforeLoan\": 590.95,\n",
    "        \"PreviousEarlyRepaymentsBefoleLoan\": 0.0,\n",
    "        \"PreviousEarlyRepaymentsCountBeforeLoan\": 0.0,\n",
    "        \"Council_house\": False,\n",
    "        \"Homeless\": False,\n",
    "        \"Joint_ownership\": False,\n",
    "        \"Joint_tenant\": False,\n",
    "        \"Living_with_parents\": False,\n",
    "        \"Mortgage\": False,\n",
    "        \"Other\": False,\n",
    "        \"Owner\": False,\n",
    "        \"Owner_with_encumbrance\": True,\n",
    "        \"Tenant\": True,\n",
    "        \"Entrepreneur\": False,\n",
    "        \"Fully\": False,\n",
    "        \"Partially\": False,\n",
    "        \"Retiree\": True,\n",
    "        \"Self_employed\": False\n",
    "}\n",
    "\n",
    "feature_df = pd.DataFrame([feature_values])\n",
    "feature_df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9e6460fb",
   "metadata": {},
   "source": [
    "We can see that this application will be rejected with a probability of $\\sim77\\%$:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "6b9377a3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0.7835957  0.21640435]]\n",
      "Paid loan is predicted as: [False]\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "\n",
    "print(xg_model.predict_proba(feature_df))\n",
    "print(f\"Paid loan is predicted as: {xg_model.predict(feature_df)}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3c3486cf",
   "metadata": {},
   "source": [
    "We will now prepare the XGBoost model to be used from the TrustyAI. Since our model outputs a Numpy array, we need to specify the column names of the output. Additionally, we must make sure our model's outputs are of shape `[n_inputs, n_outputs]`, which we can do by wrapping the model output into a DataFrame:"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7edea373",
   "metadata": {},
   "source": [
    "Once the prediction method is created, we wrap in a `Model` class.\n",
    "\n",
    "This class takes care of all the JVM's asynchronous plumbing for us."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "3cb72214",
   "metadata": {},
   "outputs": [],
   "source": [
    "from trustyai.model import Model\n",
    "\n",
    "model = Model(xg_model.predict, output_names=['PaidLoan'])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b3c362f5",
   "metadata": {},
   "source": [
    "We can confirm now, with the newly created `PredictionProvider` model that this input will lead to a `false` `PaidLoan` prediction:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "a2a997d2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([False])"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "output = model(feature_df)\n",
    "output"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "925d40d2",
   "metadata": {},
   "source": [
    "We execute the LIME explainer on the XGBoost model and prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "d1c99d0f",
   "metadata": {},
   "outputs": [],
   "source": [
    "from trustyai.explainers import LimeExplainer\n",
    "\n",
    "explainer = LimeExplainer(samples=100, perturbations=2, seed=23, normalise_weights=False)\n",
    "explanation = explainer.explain(inputs=feature_df, outputs=output, model=model)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "402ac28a",
   "metadata": {},
   "source": [
    "We output the top 2 most important features for the prediction outcome"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "d906c8b8",
   "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>Feature</th>\n",
       "      <th>Value</th>\n",
       "      <th>Saliency</th>\n",
       "      <th>Confidence</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Fully</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-9.039604e-10</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Owner</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-5.065592e-10</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Homeless</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-3.075532e-10</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>IncomeFromChildSupport</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.827385e-10</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>LoanDuration</td>\n",
       "      <td>60.0</td>\n",
       "      <td>5.266562e-10</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Self_employed</td>\n",
       "      <td>0.0</td>\n",
       "      <td>6.746940e-10</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  Feature  Value      Saliency  Confidence\n",
       "0                   Fully    0.0 -9.039604e-10         0.0\n",
       "2                   Owner    0.0 -5.065592e-10         0.0\n",
       "5                Homeless    0.0 -3.075532e-10         0.0\n",
       "1  IncomeFromChildSupport    0.0  2.827385e-10         0.0\n",
       "4            LoanDuration   60.0  5.266562e-10         0.0\n",
       "3           Self_employed    0.0  6.746940e-10         0.0"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "explanation.as_dataframe()['PaidLoan'].sort_values(by=\"Saliency\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "af99031e",
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "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>Feature</th>\n",
       "      <th>Value</th>\n",
       "      <th>Saliency</th>\n",
       "      <th>Confidence</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Fully</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-9.039604e-10</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>IncomeFromChildSupport</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.827385e-10</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Owner</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-5.065592e-10</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Self_employed</td>\n",
       "      <td>0.0</td>\n",
       "      <td>6.746940e-10</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>LoanDuration</td>\n",
       "      <td>60.0</td>\n",
       "      <td>5.266562e-10</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Homeless</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-3.075532e-10</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  Feature  Value      Saliency  Confidence\n",
       "0                   Fully    0.0 -9.039604e-10         0.0\n",
       "1  IncomeFromChildSupport    0.0  2.827385e-10         0.0\n",
       "2                   Owner    0.0 -5.065592e-10         0.0\n",
       "3           Self_employed    0.0  6.746940e-10         0.0\n",
       "4            LoanDuration   60.0  5.266562e-10         0.0\n",
       "5                Homeless    0.0 -3.075532e-10         0.0"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "explanation.as_dataframe()['PaidLoan']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "04c0d22b",
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1600x900 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "explanation.plot('PaidLoan')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.11.7"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}