feature_engineering/06-target-encoding.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "4d243c6f",
   "metadata": {
    "papermill": {
     "duration": 0.01845,
     "end_time": "2022-04-10T23:33:53.487555",
     "exception": false,
     "start_time": "2022-04-10T23:33:53.469105",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "**This notebook is an exercise in the [Feature Engineering](https://www.kaggle.com/learn/feature-engineering) course.  You can reference the tutorial at [this link](https://www.kaggle.com/ryanholbrook/target-encoding).**\n",
    "\n",
    "---\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "22fc8501",
   "metadata": {
    "papermill": {
     "duration": 0.017111,
     "end_time": "2022-04-10T23:33:53.522869",
     "exception": false,
     "start_time": "2022-04-10T23:33:53.505758",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Introduction #\n",
    "\n",
    "In this exercise, you'll apply target encoding to features in the [*Ames*](https://www.kaggle.com/c/house-prices-advanced-regression-techniques/data) dataset.\n",
    "\n",
    "Run this cell to set everything up!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "9c6e3e04",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-10T23:33:53.571506Z",
     "iopub.status.busy": "2022-04-10T23:33:53.571001Z",
     "iopub.status.idle": "2022-04-10T23:33:55.492408Z",
     "shell.execute_reply": "2022-04-10T23:33:55.491874Z",
     "shell.execute_reply.started": "2022-04-10T23:14:47.298969Z"
    },
    "papermill": {
     "duration": 1.951684,
     "end_time": "2022-04-10T23:33:55.492533",
     "exception": false,
     "start_time": "2022-04-10T23:33:53.540849",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Setup feedback system\n",
    "from learntools.core import binder\n",
    "binder.bind(globals())\n",
    "from learntools.feature_engineering_new.ex6 import *\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import warnings\n",
    "from category_encoders import MEstimateEncoder\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from xgboost import XGBRegressor\n",
    "\n",
    "# Set Matplotlib defaults\n",
    "plt.style.use(\"seaborn-whitegrid\")\n",
    "plt.rc(\"figure\", autolayout=True)\n",
    "plt.rc(\n",
    "    \"axes\",\n",
    "    labelweight=\"bold\",\n",
    "    labelsize=\"large\",\n",
    "    titleweight=\"bold\",\n",
    "    titlesize=14,\n",
    "    titlepad=10,\n",
    ")\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "\n",
    "def score_dataset(X, y, model=XGBRegressor()):\n",
    "    # Label encoding for categoricals\n",
    "    for colname in X.select_dtypes([\"category\", \"object\"]):\n",
    "        X[colname], _ = X[colname].factorize()\n",
    "    # Metric for Housing competition is RMSLE (Root Mean Squared Log Error)\n",
    "    score = cross_val_score(\n",
    "        model, X, y, cv=5, scoring=\"neg_mean_squared_log_error\",\n",
    "    )\n",
    "    score = -1 * score.mean()\n",
    "    score = np.sqrt(score)\n",
    "    return score\n",
    "\n",
    "\n",
    "df = pd.read_csv(\"../input/fe-course-data/ames.csv\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e63db2e1",
   "metadata": {
    "papermill": {
     "duration": 0.018031,
     "end_time": "2022-04-10T23:33:55.529033",
     "exception": false,
     "start_time": "2022-04-10T23:33:55.511002",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "-------------------------------------------------------------------------------\n",
    "\n",
    "First you'll need to choose which features you want to apply a target encoding to. Categorical features with a large number of categories are often good candidates. Run this cell to see how many categories each categorical feature in the *Ames* dataset has."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "6a0a36ff",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-10T23:33:55.572179Z",
     "iopub.status.busy": "2022-04-10T23:33:55.571651Z",
     "iopub.status.idle": "2022-04-10T23:33:55.610875Z",
     "shell.execute_reply": "2022-04-10T23:33:55.611291Z",
     "shell.execute_reply.started": "2022-04-10T23:14:49.492393Z"
    },
    "papermill": {
     "duration": 0.064232,
     "end_time": "2022-04-10T23:33:55.611458",
     "exception": false,
     "start_time": "2022-04-10T23:33:55.547226",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MSSubClass       16\n",
       "MSZoning          7\n",
       "Street            2\n",
       "Alley             3\n",
       "LotShape          4\n",
       "LandContour       4\n",
       "Utilities         3\n",
       "LotConfig         5\n",
       "LandSlope         3\n",
       "Neighborhood     28\n",
       "Condition1        9\n",
       "Condition2        8\n",
       "BldgType          5\n",
       "HouseStyle        8\n",
       "OverallQual      10\n",
       "OverallCond       9\n",
       "RoofStyle         6\n",
       "RoofMatl          8\n",
       "Exterior1st      16\n",
       "Exterior2nd      17\n",
       "MasVnrType        5\n",
       "ExterQual         4\n",
       "ExterCond         5\n",
       "Foundation        6\n",
       "BsmtQual          6\n",
       "BsmtCond          6\n",
       "BsmtExposure      5\n",
       "BsmtFinType1      7\n",
       "BsmtFinType2      7\n",
       "Heating           6\n",
       "HeatingQC         5\n",
       "CentralAir        2\n",
       "Electrical        6\n",
       "KitchenQual       5\n",
       "Functional        8\n",
       "FireplaceQu       6\n",
       "GarageType        7\n",
       "GarageFinish      4\n",
       "GarageQual        6\n",
       "GarageCond        6\n",
       "PavedDrive        3\n",
       "PoolQC            5\n",
       "Fence             5\n",
       "MiscFeature       6\n",
       "SaleType         10\n",
       "SaleCondition     6\n",
       "dtype: int64"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.select_dtypes([\"object\"]).nunique()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b1840d9b",
   "metadata": {
    "papermill": {
     "duration": 0.018588,
     "end_time": "2022-04-10T23:33:55.649027",
     "exception": false,
     "start_time": "2022-04-10T23:33:55.630439",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "We talked about how the M-estimate encoding uses smoothing to improve estimates for rare categories. To see how many times a category occurs in the dataset, you can use the `value_counts` method. This cell shows the counts for `SaleType`, but you might want to consider others as well."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "5863805e",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-10T23:33:55.690188Z",
     "iopub.status.busy": "2022-04-10T23:33:55.689650Z",
     "iopub.status.idle": "2022-04-10T23:33:55.701546Z",
     "shell.execute_reply": "2022-04-10T23:33:55.700981Z",
     "shell.execute_reply.started": "2022-04-10T23:16:07.853622Z"
    },
    "papermill": {
     "duration": 0.034066,
     "end_time": "2022-04-10T23:33:55.701661",
     "exception": false,
     "start_time": "2022-04-10T23:33:55.667595",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "WD       2536\n",
       "New       239\n",
       "COD        87\n",
       "ConLD      26\n",
       "CWD        12\n",
       "ConLI       9\n",
       "ConLw       8\n",
       "Oth         7\n",
       "Con         5\n",
       "VWD         1\n",
       "Name: SaleType, dtype: int64"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[\"SaleType\"].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1f3f28ed",
   "metadata": {
    "papermill": {
     "duration": 0.019216,
     "end_time": "2022-04-10T23:33:55.740682",
     "exception": false,
     "start_time": "2022-04-10T23:33:55.721466",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# 1) Choose Features for Encoding\n",
    "\n",
    "Which features did you identify for target encoding? After you've thought about your answer, run the next cell for some discussion."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "a73e4ba1",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-10T23:33:55.786176Z",
     "iopub.status.busy": "2022-04-10T23:33:55.785627Z",
     "iopub.status.idle": "2022-04-10T23:33:55.790397Z",
     "shell.execute_reply": "2022-04-10T23:33:55.790916Z",
     "shell.execute_reply.started": "2022-04-10T23:16:30.733255Z"
    },
    "papermill": {
     "duration": 0.030767,
     "end_time": "2022-04-10T23:33:55.791091",
     "exception": false,
     "start_time": "2022-04-10T23:33:55.760324",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "application/javascript": [
       "parent.postMessage({\"jupyterEvent\": \"custom.exercise_interaction\", \"data\": {\"outcomeType\": 1, \"valueTowardsCompletion\": 0.3333333333333333, \"interactionType\": 1, \"questionType\": 4, \"questionId\": \"1_Q1\", \"learnToolsVersion\": \"0.3.4\", \"failureMessage\": \"\", \"exceptionClass\": \"\", \"trace\": \"\"}}, \"*\")"
      ],
      "text/plain": [
       "<IPython.core.display.Javascript object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "<span style=\"color:#33cc33\">Correct:</span> \n",
       "\n",
       " The `Neighborhood` feature looks promising. It has the most categories of any feature, and several categories are rare. Others that could be worth considering are `SaleType`, `MSSubClass`, `Exterior1st`, `Exterior2nd`. In fact, almost any of the nominal features would be worth trying because of the prevalence of rare categories."
      ],
      "text/plain": [
       "Correct: \n",
       "\n",
       " The `Neighborhood` feature looks promising. It has the most categories of any feature, and several categories are rare. Others that could be worth considering are `SaleType`, `MSSubClass`, `Exterior1st`, `Exterior2nd`. In fact, almost any of the nominal features would be worth trying because of the prevalence of rare categories."
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# View the solution (Run this cell to receive credit!)\n",
    "q_1.check()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "164a344e",
   "metadata": {
    "papermill": {
     "duration": 0.020054,
     "end_time": "2022-04-10T23:33:55.832557",
     "exception": false,
     "start_time": "2022-04-10T23:33:55.812503",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "-------------------------------------------------------------------------------\n",
    "\n",
    "Now you'll apply a target encoding to your choice of feature. As we discussed in the tutorial, to avoid overfitting, we need to fit the encoder on data heldout from the training set. Run this cell to create the encoding and training splits:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "0e2be068",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-10T23:33:55.885525Z",
     "iopub.status.busy": "2022-04-10T23:33:55.876217Z",
     "iopub.status.idle": "2022-04-10T23:33:55.890328Z",
     "shell.execute_reply": "2022-04-10T23:33:55.890998Z",
     "shell.execute_reply.started": "2022-04-10T23:18:15.989925Z"
    },
    "lines_to_next_cell": 2,
    "papermill": {
     "duration": 0.03802,
     "end_time": "2022-04-10T23:33:55.891197",
     "exception": false,
     "start_time": "2022-04-10T23:33:55.853177",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Encoding split\n",
    "X_encode = df.sample(frac=0.20, random_state=0)\n",
    "y_encode = X_encode.pop(\"SalePrice\")\n",
    "\n",
    "# Training split\n",
    "X_pretrain = df.drop(X_encode.index)\n",
    "y_train = X_pretrain.pop(\"SalePrice\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6827ecfb",
   "metadata": {
    "papermill": {
     "duration": 0.030031,
     "end_time": "2022-04-10T23:33:55.951598",
     "exception": false,
     "start_time": "2022-04-10T23:33:55.921567",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# 2) Apply M-Estimate Encoding\n",
    "\n",
    "Apply a target encoding to your choice of categorical features. Also choose a value for the smoothing parameter `m` (any value is okay for a correct answer)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "2247e4d0",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-10T23:33:56.018790Z",
     "iopub.status.busy": "2022-04-10T23:33:56.014605Z",
     "iopub.status.idle": "2022-04-10T23:33:56.119375Z",
     "shell.execute_reply": "2022-04-10T23:33:56.119894Z",
     "shell.execute_reply.started": "2022-04-10T23:21:08.103891Z"
    },
    "lines_to_next_cell": 0,
    "papermill": {
     "duration": 0.1382,
     "end_time": "2022-04-10T23:33:56.120054",
     "exception": false,
     "start_time": "2022-04-10T23:33:55.981854",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "application/javascript": [
       "parent.postMessage({\"jupyterEvent\": \"custom.exercise_interaction\", \"data\": {\"outcomeType\": 1, \"valueTowardsCompletion\": 0.3333333333333333, \"interactionType\": 1, \"questionType\": 2, \"questionId\": \"2_Q2\", \"learnToolsVersion\": \"0.3.4\", \"failureMessage\": \"\", \"exceptionClass\": \"\", \"trace\": \"\"}}, \"*\")"
      ],
      "text/plain": [
       "<IPython.core.display.Javascript object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "<span style=\"color:#33cc33\">Correct</span>"
      ],
      "text/plain": [
       "Correct"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# YOUR CODE HERE: Create the MEstimateEncoder\n",
    "# Choose a set of features to encode and a value for m\n",
    "encoder = MEstimateEncoder(cols=['Neighborhood', 'SaleType'], m=10)\n",
    "\n",
    "\n",
    "# Fit the encoder on the encoding split\n",
    "encoder.fit(X_encode, y_encode)\n",
    "\n",
    "# Encode the training split\n",
    "X_train = encoder.transform(X_pretrain, y_train)\n",
    "\n",
    "\n",
    "# Check your answer\n",
    "q_2.check()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "f8204877",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-10T23:33:56.187655Z",
     "iopub.status.busy": "2022-04-10T23:33:56.186589Z",
     "iopub.status.idle": "2022-04-10T23:33:56.190668Z",
     "shell.execute_reply": "2022-04-10T23:33:56.190138Z",
     "shell.execute_reply.started": "2022-04-10T23:21:37.260445Z"
    },
    "lines_to_next_cell": 0,
    "papermill": {
     "duration": 0.038786,
     "end_time": "2022-04-10T23:33:56.190793",
     "exception": false,
     "start_time": "2022-04-10T23:33:56.152007",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Lines below will give you a hint or solution code\n",
    "#q_2.hint()\n",
    "#q_2.solution()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1bb0ad7c",
   "metadata": {
    "papermill": {
     "duration": 0.031656,
     "end_time": "2022-04-10T23:33:56.255702",
     "exception": false,
     "start_time": "2022-04-10T23:33:56.224046",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "If you'd like to see how the encoded feature compares to the target, you can run this cell:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "ceb321e7",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-10T23:33:56.328453Z",
     "iopub.status.busy": "2022-04-10T23:33:56.327603Z",
     "iopub.status.idle": "2022-04-10T23:33:56.740778Z",
     "shell.execute_reply": "2022-04-10T23:33:56.740200Z",
     "shell.execute_reply.started": "2022-04-10T23:21:50.430515Z"
    },
    "papermill": {
     "duration": 0.453581,
     "end_time": "2022-04-10T23:33:56.740918",
     "exception": false,
     "start_time": "2022-04-10T23:33:56.287337",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 540x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "feature = encoder.cols\n",
    "\n",
    "plt.figure(dpi=90)\n",
    "ax = sns.distplot(y_train, kde=True, hist=False)\n",
    "ax = sns.distplot(X_train[feature], color='r', ax=ax, hist=True, kde=False, norm_hist=True)\n",
    "ax.set_xlabel(\"SalePrice\");"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "47d64c22",
   "metadata": {
    "papermill": {
     "duration": 0.033575,
     "end_time": "2022-04-10T23:33:56.807996",
     "exception": false,
     "start_time": "2022-04-10T23:33:56.774421",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "From the distribution plots, does it seem like the encoding is informative?\n",
    "\n",
    "And this cell will show you the score of the encoded set compared to the original set:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "a927e30e",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-10T23:33:56.879540Z",
     "iopub.status.busy": "2022-04-10T23:33:56.876379Z",
     "iopub.status.idle": "2022-04-10T23:34:04.381426Z",
     "shell.execute_reply": "2022-04-10T23:34:04.382051Z",
     "shell.execute_reply.started": "2022-04-10T23:23:12.727195Z"
    },
    "papermill": {
     "duration": 7.541201,
     "end_time": "2022-04-10T23:34:04.382231",
     "exception": false,
     "start_time": "2022-04-10T23:33:56.841030",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Baseline Score: 0.1428 RMSLE\n",
      "Score with Encoding: 0.1399 RMSLE\n"
     ]
    }
   ],
   "source": [
    "X = df.copy()\n",
    "y = X.pop(\"SalePrice\")\n",
    "score_base = score_dataset(X, y)\n",
    "score_new = score_dataset(X_train, y_train)\n",
    "\n",
    "print(f\"Baseline Score: {score_base:.4f} RMSLE\")\n",
    "print(f\"Score with Encoding: {score_new:.4f} RMSLE\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e328de4d",
   "metadata": {
    "papermill": {
     "duration": 0.022109,
     "end_time": "2022-04-10T23:34:04.426824",
     "exception": false,
     "start_time": "2022-04-10T23:34:04.404715",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Do you think that target encoding was worthwhile in this case? Depending on which feature or features you chose, you may have ended up with a score significantly worse than the baseline. In that case, it's likely the extra information gained by the encoding couldn't make up for the loss of data used for the encoding."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "717b0254",
   "metadata": {
    "papermill": {
     "duration": 0.022261,
     "end_time": "2022-04-10T23:34:04.471183",
     "exception": false,
     "start_time": "2022-04-10T23:34:04.448922",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "-------------------------------------------------------------------------------\n",
    "\n",
    "In this question, you'll explore the problem of overfitting with target encodings. This will illustrate this importance of training fitting target encoders on data held-out from the training set.\n",
    "\n",
    "So let's see what happens when we fit the encoder and the model on the *same* dataset. To emphasize how dramatic the overfitting can be, we'll mean-encode a feature that should have no relationship with `SalePrice`, a count: `0, 1, 2, 3, 4, 5, ...`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "61044a72",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-10T23:34:04.518971Z",
     "iopub.status.busy": "2022-04-10T23:34:04.518436Z",
     "iopub.status.idle": "2022-04-10T23:34:07.951353Z",
     "shell.execute_reply": "2022-04-10T23:34:07.951969Z",
     "shell.execute_reply.started": "2022-04-10T23:30:29.429138Z"
    },
    "papermill": {
     "duration": 3.459023,
     "end_time": "2022-04-10T23:34:07.952147",
     "exception": false,
     "start_time": "2022-04-10T23:34:04.493124",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Score: 0.0293 RMSLE\n"
     ]
    }
   ],
   "source": [
    "# Try experimenting with the smoothing parameter m\n",
    "# Try 0, 1, 5, 50\n",
    "m = 0\n",
    "\n",
    "X = df.copy()\n",
    "y = X.pop('SalePrice')\n",
    "\n",
    "# Create an uninformative feature\n",
    "X[\"Count\"] = range(len(X))\n",
    "X[\"Count\"][1] = 0  # actually need one duplicate value to circumvent error-checking in MEstimateEncoder\n",
    "\n",
    "# fit and transform on the same dataset\n",
    "encoder = MEstimateEncoder(cols=\"Count\", m=m)\n",
    "X = encoder.fit_transform(X, y)\n",
    "\n",
    "# Results\n",
    "score =  score_dataset(X, y)\n",
    "print(f\"Score: {score:.4f} RMSLE\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3482424d",
   "metadata": {
    "papermill": {
     "duration": 0.022761,
     "end_time": "2022-04-10T23:34:07.998726",
     "exception": false,
     "start_time": "2022-04-10T23:34:07.975965",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Almost a perfect score!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "fff61dd7",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-10T23:34:08.050650Z",
     "iopub.status.busy": "2022-04-10T23:34:08.050120Z",
     "iopub.status.idle": "2022-04-10T23:34:08.418936Z",
     "shell.execute_reply": "2022-04-10T23:34:08.418530Z",
     "shell.execute_reply.started": "2022-04-10T23:30:33.789310Z"
    },
    "papermill": {
     "duration": 0.397583,
     "end_time": "2022-04-10T23:34:08.419047",
     "exception": false,
     "start_time": "2022-04-10T23:34:08.021464",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAhIAAAFfCAYAAAAbEyWQAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAA3XAAAN1wFCKJt4AABBiElEQVR4nO3dd3iV9f3/8ed9Ts45OZkQEkbYIgGUvSJbqeLWgrv+kJZWqrhqW620aqtttfXrrODAilixWhUp7lppRZEdGQbZmwAhg+zkzPv3x0mOUFYSktznJK/HdeUiOffJfd45N8l5nc80TNM0EREREakHm9UFiIiISPRSkBAREZF6U5AQERGRelOQEBERkXpTkBAREZF6U5AQERGRelOQEBERkXpTkBAREZF6U5AQERGReouqIDFv3jwmTZpE3759mT59eqM9TlZWFtdffz2DBg1ixIgRPPPMM432WCIiItEsxuoC6qJt27ZMnz6dpUuXcvDgwUZ5jE2bNnH77bfz+9//nrFjx+L3+9mzZ0+jPJaIiEi0i6ogMWHCBAA2btx4TJAoKCjgkUceYcWKFQBcfPHF3HPPPTidzjo9xnPPPcc111zD+eefD4DT6aR3794NUL2IiEjzE1VdGydimia33noraWlp/Pvf/+b9999n06ZNPPfcc3U+16pVq/D5fFx55ZWcc845/PjHP2bHjh2NULWIiEj0axZB4ptvvmH37t3ce++9uN1uWrduzS233MIHH3wAQGFhIb169Trhx+TJk8PnKi4u5sMPP+T//u//+OKLL+jduzfTp0/H7/db9eOJiIhErKjq2jiRnJwcSkpKGD58ePg20zQJBoMAtG7dmmXLlp3w+x0OR/jzuLg4Jk2aREZGBgB33XUXr7zyCrt27eLMM89spJ9AREQkOjWLINGhQwfatGnDkiVLjnvcMAxSUlJqdS6NhxAREam9qOra8Pv9eDwe/H4/wWAQj8eD1+ulX79+tG/fnqeeeoqysjJM0yQnJ4fFixfX+TGuvfZaFixYwI4dO/D5fMyaNYuuXbvSrVu3hv+BREREolxUtUg8//zzzJw5M/x1//79GT58OK+99hovvvgijz/+OJdccgllZWWkp6dz3XXX1fkxrrjiCg4ePMhNN92Ex+Ohf//+PP/888TERNVTJSIi0iQM0zRNq4sQERGR6BRVXRsiIiISWRQkREREpN4ivuM/GAxSXl6Ow+HAMAyryxEREWkxTNPE5/MRHx+PzXb8toeIDxLl5eVs2bLF6jJERERarIyMDBITE497LOKDRM1iURkZGXXeN6MpZWdn07dvX6vLkBPQ9Ylsuj6RTdcnsjXm9fF6vWzZsuWohRv/V8QHiZruDKfTicvlsriak4v0+lo6XZ/IpusT2XR9IltjX5+TDS3QYEsRERGpNwUJERERqTcFCREREak3BQkRERGpNwUJERERqTcFCREREak3BQkRERGpNwUJERERqTcFCREREak3BQkRERGpt4hfIlui1OzZJz8+bVrT1CEiIo1KLRIiIiJSbwoSIiIiUm8KEiIiIlJvChIiIiJSbwoSIiIiUm8KEiIiIlJvChIiIiJSbwoSIiIiUm8KEiIiIlJvChIiIiJSbwoSIiIiUm8KEiIiIlJvChIiIiJSbwoSIiIiUm8KEiIiIlJvlgaJ3Nxcpk+fTmZmJpmZmdx1110UFhZaWZKIiIjUgaVB4qGHHgLgP//5D4sWLcLj8fCHP/zBypJERESkDiwNEnv37uXiiy8mPj6ehIQELrnkErZs2WJlSSIiIlIHlgaJH/3oR3zyySeUlpZSUlLChx9+yHnnnWdlSSIiIlIHhmmaplUPvmvXLu677z7Wrl0LwMCBA/nrX/9KQkJC+D4ej4fs7GyLKpT6Sn333ZMez580qYkqERGR09W3b19cLtdxj8U0cS1hwWCQqVOnctFFF/HKK68A8OyzzzJ16lTeeuutY+5/sh8iEmRlZTFkyBCry4gcWVknPdy1iZ8rXZ/IpusT2XR9IltjXp/avJm3rGujqKiInJwcbrrpJtxuN263m8mTJ7Nu3TrN3BAREYkSlgWJlJQUunbtyuuvv47H48Hj8fD666/Tvn17UlJSrCpLRERE6sDSwZbPPfcc3377LWPHjmX06NGsX7+e559/3sqSREREpA4sGyMBcOaZZ/Lyyy9bWYKIiIicBkuDhESx2bOtrkBERCKA9toQERGRelOQEBERkXpTkBAREZF6U5AQERGRelOQEBERkXpTkBAREZF6U5AQERGRetM6EtKoXi2OZ35ZHGVBA7dh8kK7Qjo7AlaXJSIiDURBQhrNNm8MDxckE8AI3/aLvNa82SFfTWEiIs2E/p5Lo/lTYRIBDH7Zupj13fZzltPLyioXc4rjrS5NREQaiIKENIqvKl18VuGmc4yfm1uVkWQzebLtYRyYPHY4mW2HyqwuUUREGoCChDQ404RHCpIAmJFSjKu6Z6O3089drUvwmgYvL9lhYYUiItJQFCSkwW33xbDB66SP08vF8VVHHZucVI7LMHl/3QEqvRp0KSIS7RQkpMF9XhELwAVxVRjG0ceS7SYXxlVS5vHzrw0HLahOREQakoKENLjPK10AjIvzHPf41YkVALydtbfJahIRkcahICENqiJosLLSRbItyECX97j3GeX20CE5lqXbC9h3uKKJKxQRkYakICENalmlCy8GY9xV2I3j38duwKTBHTFNePfrnKYtUEREGpSChDSoxafo1qgxcVAnAI2TEBGJcgoS0mBM87uBluPcVSe9b4+0eLq2iWPD/hJyS05+XxERiVwKEtJgdvvt7PHHcJbTS9uY4EnvaxgG5/VqC8Dnmw81RXkiItIItNeGNJh1HicAw2OPP8jyKLNnc16Fi7mk8t+PV3Ldmk+OPj5tWiNUKCIiDU0tEtJgNngcAPR1+Wp1/8xYD24jyJJKF16zMSsTEZHGoiAhDSY7HCRq0SIBxNpCU0HLTBurq5yNWZqIiDQSBQlpEKYJ2V4nLsPkTIe/1t93blxooOV/qgdpiohIdFGQkAaxz2+nJGijt9NHzAnWjzie86qniX6uICEiEpUUJKRBbPBWd2s4a9etUaNjTIAzHD62+Rwc9Ou/o4hItNFfbmkQ2XUcaHmkUe5Qq8TS6sWsREQkeihISIPIrp76eTpB4isFCRGRqKMgIactNNDSQQwmGc66B4kRsR5smCytdGFqGqiISFRRkJDTdihgIz9gp6fTh6sOAy1rJNtN+rp8HAjEsNOnNdJERKKJgoSctnC3Rj1aI2qMjFX3hohINFKQkNO2pboVoXc9xkfUGK1xEiIiUUlBQk7bdm8oSNRlIar/NTTWg9MwWVblIqBxEiIiUUNBQk7bdl9o6meP0wgSsTYY4vJSHLTxbfWaFCIiEvkUJOS0mCZs98UQawRJjwmc1rlGuUPLZat7Q0QkeihIyGkpCNooCdro7vBjq8eMjSON1DgJEZGooyAhp6VmfMTpdGvU6O/ykWgEWVXlxOM/vdYNERFpGgoSclq2+xouSMQYkOn2UGXa+Hp30WmfT0REGp+ChJyW8EBL5+kHCThi343t+Q1yPhERaVwKEnJaaro2zmiAFgk4Yt+NbQoSIiLRQEFCTssOX8MGiZ4OP2n2AOv2FVNaVf8FrkREpGkoSEi9VQVhr99Oxxg/cbaGWUXKMEKtEoGgyYodhQ1yThERaTwKElJvu/wxmBgN1hpRIzwNVOMkREQinoKE1FtDTv08UnjA5baCBj2viIg0PAUJqbcdDbA09vF0jAnQrU0cm3NLOVRa1aDnFhGRhqUgIfW2s3qgZfcGDhIAI89MBWDZdrVKiIhEMgUJqbfdPjsAXRshSIyuDhKaBioiEtkUJKTedvtjcGCe9mZdxzPijDYYBny1rQDT1L7iIiKRSkFC6qU8aJAfsNPJEcB+mpt1HU/reCdndUgip6iSPYUVDf8AIiLSIBQkpF72VHdrdIlp+G6NGqPC3RsaJyEiEqkUJKRedvtDAy0bY3xEjZE92gAaJyEiEsksDxKLFi3iyiuvZODAgYwePZo33njD6pKkFvZUz9hozBaJ4d1TcNgNlm7PJxjUOAkRkUgUY+WDf/HFFzz00EP83//9H0OHDqWsrIz8fL37jAbfzdho+IGWNeKcMQzq0pqVOwvZeLCEs9OTG+2xRESkfixtkXjmmWe47bbbyMzMxG63k5ycTI8ePawsSWppTxN0bQCM6hEaJ6FVLkVEIpNlQaKiooINGzaQm5vLhRdeyKhRo7jzzjs5dOiQVSVJHexugq4NgFFnVo+T0L4bIiIRyTAtmqR/8OBBxo0bR69evXj++edp1aoVv/3tb8nLy+PVV18N38/j8ZCdnW1FiXIC/qDJD+YfpDVe5tuzGuUx8idNCj/WlIWHME149fttcdgaYa6piIicVN++fXG5XMc9ZtkYibi4OAAmT55Mx44dAbjzzjuZMGECFRUV4eM1TvZDRIKsrCyGDBlidRlNYndBOQFyOSMWuqZ3bZTH6HrEczlywyr+s+kQttQzGNI9pV7na0nXJxrp+kQ2XZ/I1pjXpzZv5i3r2khKSiI9Pf24x7SSYWTbXRBaIKqxuzVqaBqoiEjksnSw5bXXXsu8efPIzc2lqqqKWbNmMWLECOLj460sS05hd/VKk405Y+NINQtTLdU4CRGRiGPp9M9p06ZRXFzMFVdcAUBmZiaPPfaYlSVJLewpKAegSyPP2KjRq10ibeKdrNlTRLnHT7zL0v+2IiJyBEv/Itvtdu677z7uu+8+K8uQOqrp2ujaRF0bNpvByDNTeX/dflbuLOS83m2b5HFFROTULF/ZUqJPzSZaXZqoawNglMZJiIhEJAUJqRPTNNl3uJIEI0hrW7DJHje8gdd2LUwlIhJJFCSkTooqfJR5/HRyBDCacEmHzilxdE5xs/FACQVlnqZ7YBEROSkFCamTvYdD3Rqdm2h8xJFqlstetkOtEiIikUJBQupkb2ElAJ1imm58RI1w94bGSYiIRAwFCamTcItEE039PFLNwlRLFCRERCKGJuRLnewtrOnaaOQWidmzj7mpDdDHmcbGQthTUEGXNnHHfp+IiDQptUhInew9HOrasKJFAmCMOzTQUq0SIiKRQUFC6mRfU7VInMCo6iChcRIiIpFBXRtSa8FgaA2JlHgn8TZrNlYbHuvFiclXG3IIvLga+/GmoE6b1uR1iYi0VHVukQgErHknKtY7VOrBGwjSubXbshrcNpMhsV6Kgja+9Tosq0NERELqHCTGjBnDH//4x1PuTy7NT82MjU4p1g5yHO2uAuDLSpeldYiISD2CRGFhIfPmzeOaa67h0ksvZfbs2Rw8eLAxapMIE56x0drqIFE9TkJBQkTEcnUOEpdeeinx8fGYpsn27dt56qmnGD9+PFOmTGHBggWUl5c3Rp0SAWoWo+qcYl3XBkBfl49kW5BVVS6qmm67DxEROY46B4knnniCZcuW8eKLLzJp0iSSk5MJBoOsXLmSX//614wePZoZM2awdevWxqhXLBRejMriFgm7ASPdHrymwaoqtUqIiFipXtM/HQ4H48aN45FHHuGNN95g8ODBmGZoFH9lZSULFixg4sSJfPLJJw1arFgr3LVh8RgJgFHV4ySWqHtDRMRS9Zr+WVVVxSeffML8+fPJysoKhwjDMBg3bhyHDx9m3bp1PPXUU1x00UUNWrBYZ9/hSgwD0lvFWl3KdwtTKUiIiFiqzkHigQce4OOPP6a8vDwcIFJTU7nqqqu47rrrSE9PJxAIMGrUKPbt29fgBYs1fIEgB4oraZcYiyvGbnU5dIkJ0CnGzwavk8KAjRS7BkuIiFihzkHi7bffDn8+bNgwbrjhBiZMmEBMzHenstvt9OrVi7179zZMlWK5/UWVBE3oEgHdGgCGEWqVeKM0hq8qXVyeUGl1SSIiLVKdg0RiYiLf//73uf766+nRo8cJ7/fqq6+eVmESWb6bsREZQQJCy2W/URqvICEiYqE6B4nHH3+cuLi444aIYDDUvGyzaQuP5mZPeKCltVM/jzTS7cHA5MtKF6YZaqUQEZGmVecg8dOf/hSbzca33357zLGzzjrrhMckukXK1M8jpdiDnO30ke11sttvp5tDy7eLiDS1ejUd1AyyPJLf7z/hMYl+NS0SXdpETpCA73YDXarZGyIilqhVi8TKlStZuXLlUbfNnDnzqK937twJQGys9VMDpeHti5Dlsf/XCLeHF4sTWVrp4gdJFVaXIyLS4tQ6SMyaNSv8tWmaR31dwzAMevbs2XDVScTYe7gSZ4yNtomR9c5/WKyXGEyWV2mchIiIFWrdtWGaJqZpYhgGhmGEvz7yo2vXrtx///2NWa9YoMzjp7DcS6fWbmy2yHqljreZDHR5yQ/Y2eqr1/pqIiJyGmr1l3fKlClMnDgR0zQ5//zzMQyDRYsWhY8bhkFSUhIJCQmNVqhYp2Zp7EhZQ+J/jXR7WO1xsbTSRYbTb3U5IiItSq2CRGJiIomJiQA8+uijAHTs2LHxqpKIsidCx0fUOMft4S9FoQGXP0zW7rMiIk2pVkFi//79AKSnp5OZmXnUbceTnp7eAKVJpNgbgWtIHGmwy4vTCI2TCJhg/QLeIiItR62CxPjx48PrQ4wfPx7jJCPaDMPQOhLNTKR3bcTaYKjLw9KqWL71OuhndUEiIi1InQZbHvn5yT6kedl7OLT8dKcI7dqA0DgJgGVaT0JEpEnVqkXitttuC7dC3H777Y1akESeSF2M6kgj3F44HBonMc3qYkREWpBaBYk77rgj/LmCRMtimib7DleQ7HaQFOuwupwT6u/yEm8EWVnlxBcI4rBrvxcRkaZw2n9tA4EAH3zwAX/729/YsmVLQ9QkESSvzEOVLxix4yNqOIzQ4lQVpo31+4qsLkdEpMWoc5B49NFHGTx4MH/5y1+AULfHPffcw6OPPspVV13FsmXLGrxIsU6kz9g4Us04iaXbCiyuRESk5ahzkFi1ahWVlZWMGjWKvXv38vnnn4cHWfp8Pl566aXGqFMssrcwNNAyUteQOFJ4wOUOBQkRkaZS5yCRk5MDQI8ePcjOzgZCK1/OmzcPgM2bNzdgeWK18GJUEd61AdDH6SPZFmT17sNU+bSluIhIU6hzkKisDL1DjYuLY9u2bRiGwbBhwxg4cCAAJSUlDVqgWGtvFAUJuwHnxHrw+oN8veew1eWIiLQIdQ4SqampADz11FN89NFHQKh1orCwEIDk5OQGLE+stifCF6P6XyNquje2q3tDRKQp1Hm7xJEjR/LOO+8wd+5cTNOkc+fOdO/enSVLlgBwxhlnNHiRYpHZs9m3tx0GdtLnvw6RtfHncYUHXG4v4BcW1yIi0hLUuUXi5z//OaNHj8btdtOzZ08ee+wxIDQIMz09nfHjxzd4kWINrwkH/HY62AO4oiBEAPR0+ElNcLJubxHlHu0EKiLS2OrcIpGSksJf//rXY26/++67ufvuuxukKIkM+/12ghh0ckTPwEXDgBE9Unl/3X5W7Srk3F5trS5JRKRZq3OQOFJ+fj5er/eY27X7Z/Ow1xf679E5Jrre2Y/s0Yb31+1n2fYCBQkRkUZW5yBx+PBh/vCHP/Dpp5/i9x/7AqPdP5uPPf7QhtxdoqhFAmDEGW2A0DgJERFpXHUOEg888ACfffZZY9QiESZaWyS6tokjPTmW7P3FFFf4SI6L3D1CRESiXZ2DxPLlyzEMg549ezJ27FhcLld4Z1BpXvZGaYuEYRiM6JHK/K/3sXxnARee3d7qkkREmq06Bwm32015eTlz584lJSWlMWqSCLHXH50tEhAaJzH/630s264gISLSmOo8/fO6664DYOPGjQ1ejESWPb4YnIZJmj1odSl1NqJHaJyEFqYSEWlcdW6RCAaDJCQkMH36dL73ve/RtWtX7Hb7Ufe5/fbbG6xAsUZplY+ioI0eDh+2KOy5Sm/lplubODbnlpJX6rG6HBGRZqvOQeK5554Lj4n4+OOPj3sfBYnoF971Mya6xkccaUSPVHYV7GH5jgI0IVlEpHHUuWsDCG8bfrwPaR7Cu346om98RI2RPTQNVESksdW5RWLRokWNUYdEmN0F5QB0jcKBljXOqV5PYvmOAq7ummhxNSIizVOdg0THjh0bvIiqqiouv/xyDh8+zOrVqxv8/FJ3uwpCLRLdomzq55HSEl30apfI5txS8iuiY/dSEZFoU6+ujaKiIp588kmuu+46LrroIgDef/99/vnPf1JQUPdm5GeeeUbLakeYXfmhFoluUdy1Ad/N3sg+dOxS7iIicvrqHCTy8/O56qqreOmll1i3bh27d+8GYMmSJcyYMYP58+fX6XzZ2dksWbKEm2++ua6lSCPaXVCODTOqx0jAd0Him0OauSEi0hjqHCSefvppcnJycDiOXnZ40qRJmKbJf//731qfy+/388ADD/Dggw8ecz6xTpUvwP7iKtJjomf78BM5p3sbDAO+OeTVYGARkUZQ5zESixcvxjAMXnnlFW688cbw7f369QNg7969tT7Xyy+/TJ8+fRg2bBgrVqw46X2zs7PrWmqTy8rKsrqEBrGn2AdAe39ZuMUpmuT/z3Xo0crBtsM+Fn6+ks5Jp7XhrTSi5vL701zp+kQ2K69PvXb/BBgwYMBRtweDodUPi4uLa3We3bt38+abb7JgwYJa3b9v3764XK46VNq0srKyGDJkiNVlNIj8DQeBAnonxdA1tavV5dRZ1/+5DpcUbuEvi7aSZ0/j+0POsKgqOZnm9PvTHOn6RLbGvD4ej+eUb+Tr3LWRmpoKcMxW4W+++SYAaWlptTpPVlYW+fn5XHjhhWRmZjJ9+nTKysrIzMxk3bp1dS1LGlB4oGUUT/080nm9Qv8n/7v5kMWViIg0P3VukRgzZgxvv/02t9xyS/i2SZMmsXHjRgzDYPTo0bU6z8UXX8zIkSPDX69Zs4b777+fhQsXajMwi9VM/ewexVM/j9S/UyuSnAardhVS5vGT4FL3hohIQ6lzi8Ttt99OamoqhYWF4ds2btyIaZq0adOG2267rVbncbvdtG/fPvyRkpKCYRi0b98ep9NZ17KkAdW0SHSN8hkbNew2g4HtXfgCJl9ty7e6HBGRZqXOQaJdu3bMnz+fq666irS0NOx2O23atGHixIm8/fbbtGvXrl6FZGZmajGqCLGroBybEd3LY/+vwR1C42s+35xncSUiIs1Lndt4N23axJdffonL5eKCCy4gPT2d0aNH07t378aoT5pYlS/AgeIqOrV2R/3UzyMNaOfCMODzzYcwTTO88ZyIiJyeWgeJsrIyZsyYwWeffXbMsSeeeIILLriARx55hISEhAYtUJrW7prxEanx0DyGSACQ5LIxsHMr1uwpYtPBUvp0SLK6JBGRZqFWXRs+n49p06bx2WefnXDXz3//+9/ceuutBALN6NWnBdpVvVlXtzbxFlfS8M7vE+p2+9eGgxZXIiLSfNQqSPzzn//k66+/xjRNOnXqxC9/+UtmzpzJzJkz+eUvf0mnTp0wTZPVq1fz3nvvNXbN0ojCAy3bNL9Nri48uz0A/9qQa3ElIiLNR626Nj7++GMAxo4dy6xZs45ZznrKlCnceuutLFmyhA8++ICJEyc2fKXSJHZWB4nuqc2vReLMtgn0SItn44ES9hRU0KUZhiURkaZWqxaJ7du3YxgGd99993H3xHA4HPz85z8P31ei1/a8MiD0otscfdcqoe4NEZGGUKsWiaKiIgB69ux5wvtkZGQA3y2hLdFpe145TruNTq2j+N367NnH3JS6ezdkZXGRx8FztOVfGw5y81gtly0icrpqFSS8Xi8Aa9euPeUOijX3lehTWO6lsNxLr3aJ2G3Nc3pkP6eP9ORYsvYcJq/UQ1pi5O7fIiISDWoVJGrm3U+ePLmx6xEL7aju1ujRtvmNj6hhGDDh7PbMXbqLT7IPMHlEN6tLEhGJarVe2fJE0z6P/JDotu1QdZBIa57jI2pcPqADAO+t229xJSIi0a9WLRKahdEy1Ay0bO5BYnCX1nRq7WbVrsPsO1wR3eNBREQsVqsg8eijjzZ2HRIBtueFpn429yBhGAZXDkxn1n+38966/Uw/90yrSxIRiVp13rRLmq+aFokz0prvGIkaVw7sCMDCNereEBE5HQoSAoQ269pbWEF6cizxrjrv5RZ1Mtol0rt9IptzS9l0sMTqckREolbzf8WQWtldUEHQhB7NdCGqo1SvM/F9bwJ/IpkFr3zEjDZHhIlp0ywqTEQk+qhFQoCWM9DySN9PqMCGyfyyOHyadCQiUi8KEgLA9vDUz+Y/PqJG+5gg58VVkR+ws6gi1upyRESikoKEALCtBbZIAFyXWAHAP0paToASEWlIChICwNbcmlUtW1aQOC+uijR7gMWVLvb77VaXIyISdRQkBH8gyLa8MlrFOWjbwvaecBhwdWIFQQzeLtXCVCIidaUgIewqqMDrD9KrXSKG0Tw36zqZ6xJDC3G9WRKHX4MuRUTqREFC2JJbCkCv9okWV2KNbo4A49xVHAjE8Gm5Bl2KiNSFgoSw6WDLDhIAP0wOjRGZW9KyxoiIiJwuBQlhS02QaNdyg8Q4t4fuDj8rq1x8u18rXYqI1JaChLC5umsjowW3SNgMuCkp1Crx6tJd1hYjIhJFFCRauCpfgF0F5aQnx5IU67C6HEtdnVhBvBFkwdoc8ko9VpcjIhIVFCRauK25ZZhmyx4fUSPRZvKDpHK8/iBzvtppdTkiIlFBQaKFU7fG0X6SXIbTbmPest2UVPmsLkdEJOIpSLRwm6u30O6tIAFAu5ggkwZ3pNTj5/Xle6wuR0Qk4ilItHCbq5fGzmjBMzb+17SxZ2AY8PKSnVT5AlaXIyIS0RQkWrjNB0uw24wWt1nXyZyRlsAlfTuQX+bhnax9VpcjIhLRFCRasIIyD7klHnqkxRPr0IZVR7plXA8AZn+xA38gaHE1IiKRS0GiBdtQvfDS2enJFlcSefp1SmZMz1T2FFbw4TcHrC5HRCRiKUi0YNn7iwE4Oz3J4koi063VrRLPf74d09RuXiIixxNjdQFiHbVInMDs2QCMMGGAK411B0v5z9Ov8b34Kpg2zeLiREQii1okWrCaPSXOUovEcRkG3NEqtM7Gk4cTCapRQkTkGAoSLVRplY+d+eV0TnGT7G7ZS2OfzPfiqhjg8rLB6+QTbTEuInIMBYkWauOB0DvtszuoW+NkDAPuaR1quXnycBIBNUuIiBxFQaKF2lA90LJvR3VrnMoot4fMWA/bfA4Wrs2xuhwRkYiiINFCaaBl7RkG/DIl9Hw9/dlWfFpXQkQkTEGihfouSKhFojaGxXoZ565iT2EFb6/WapciIjU0/bMF8vgDbM0tJdUeoO2bf7O6nKjxy5QSFufE8ux/tjJpcEetBioiglokWqRNB0rxB036OrVNdl30c/m48Ox2HCiu4vUV2hlURAQUJFqkdfuKABgY67W2kCj08wt6YRjw7H+2UlyhICYioiDRAq3dUwTAAJeCRF31ap/I9cM6U1Th4+lFW6wuR0TEcgoSLdDa6haJAS69o66PX0zoRYIrhteW7WbboTKryxERsZSCRAtTXOljR145XVLiSLFrGmN9pCa4uGP8mfiDJo98tNHqckRELKUg0cKsrxkf0bmVpXVEux+O6kbXNnH8Z9MhFm/Js7ocERHLKEi0MOv2FgEwQEHitLhi7Pz6kj4A/P6Db/FrkSoRaaG0jkQLs7Y6SAzsnAzfWltLVKreYhxgggkjYlNZdghef/INpiSXa5txEWlx1CLRgpimydq9xcTYDC2N3QAMAx5oU4QNk6cOJ1EY0K+TiLQ8+svXguwvriK/zEPvDolalbGBnOXyc0NiOUVBG38oUDgTkZZHQaIFWbPnMAADOrWytpBm5t6UEtraA7xbFseSrflWlyMi0qQsCxJer5f777+f8ePHM2jQIC666CLeeecdq8ppEVbvCgWJod1aW1xJ85JsN3moTREAv17wDZXegLUFiYg0IcsGW/r9ftLS0pg7dy6dO3dm3bp13HzzzbRv357Ro0dbVVbzcsTAQIDV+9IAJ0OX/gtW6cWuIV0UX8X5cZV8VgjPLNrKfRf3trokEZEmYVmLRFxcHHfddRddunTBMAwGDhxIZmYmWVlZVpXUrJUFDb71OmhnD9ApRiGioRkGPJxaTLzTzktf7uDb6m3aRUSau4gZI+HxeFi/fj29evWyupRmaW2VkyAGQ2M9GIbV1TRP6TEB7r2oN4GgyYx31xMImlaXJCLS6AzTNC3/a2eaJvfccw+5ubm8+uqr2Gzf5RuPx0N2draF1UWv1HffDX8+N9iJV8wu3GHs4GrbQQurat5yJ07kN/8pZGuhjykDErkiI97qkkRETlvfvn1xuVzHPWb5glSmafK73/2OnTt3Mnfu3KNCxJFO9kNEgqysLIYMGWJ1GUc7opto64E2UAkXpifQ1dXVwqKssXv3brp2bfyfu+vQoczsVMrlzy7hjQ3l3Dh+EBntEhv9caNdRP7+SJiuT2RrzOtTmzfzlnZtmKbJQw89xPr165kzZw6JifqD2xj8JqypchJvBOnt1I6fja1X+0R+MSEDrz/I3f9Yi9ev5bNFpPmyNEg8/PDDfP3118yZM4fkZC3m01g2eR2UmzYGxXqJ0fiIJvGTMWcwvFsKG/aX8JdFW60uR0Sk0VgWJHJycvj73//Ozp07w2tJDBo0iAcffNCqkpqtVVVOAIbGei2upOWw2wyeuHYA8U47z32+jazdh60uSUSkUVg2RqJjx45s3rzZqodvUZZXhsaWDI/1WFxJC3DE2h2dgd8mxXFvfmt+8dIXfNTpEHG33GxdbSIijSBipn9K4wiasLzKhROTwS61SDS1axIrOD+ukl3+GP5YkGR1OSIiDU5Bopn71uugOGhjcKyXWF3tJmcY8GhqEW1sAV4vTeCTbE29FZHmRS8tzVxNt8YIt7o1rJIWE+TxtqExEve+s459hyssrkhEpOEoSDRzS6tCQWKkgoSlzovzMC25lJIqPz97cy3+gKaEikjzoCDRjPlNWFnpxG0EGaDxEZb7ZUoJAzols3r3YZ7RlFARaSYUJJqxbzwOykwbQ2O9OLV+hOWcBjx7w2ASXTHM/O82lm7Lt7okEZHTpiDRjC2r0viISNOlTRyPTOqHacJd/1hLfpmujYhENwWJZmxpzUBLrR8RUS4fkM71wzqTV+rhjr+v0XgJEYlqChLNVIXXz8pKF8m2IP1c2l8j0vzuirPp0yGJZTsKePLfW6wuR0Sk3izf/VMax/IdBXgxuMBdqf01Ikn1ypexwAsxdi6zteW5z7czaONKLrj7JmtrExGpB7VINFOLN+cBcG6cujUiVVdHgCfTQutL/DyvNbsLyi2uSESk7hQkmqnFW0JBYpy7yuJK5GQuiK9ieqtSSoM2bpn3NVW+gNUliYjUiYJEM7Qrv5xdBRX0cXppG6OBfJHu561LGBlbxcYDJcx49xtM07S6JBGRWlOQaIa+2FrTGqFujWgQY8Bf2h0mPTmWBWtymPXfbVaXJCJSawoSzVDN+IhxcerWiBap9iAv/3AY8U47j3+6hQ/W77e6JBGRWlGQaGYqvQG+2p5PgiuGIbFaFjua9OmQxF9uGITNgF+8tY61e4usLklE5JQUJJqZL7bmUeULcm6vNC2LHYW+16cdv7n0LDz+ID95dTU5RZVWlyQiclJaR6KZ+deGgwBceHZ7WGFxMVI31WtMTDVhe2Ir/l4KP37yX7zZIY9WdhOmTbO4QBGRY6lFohnxBYIs2ngIp93Gub3SrC5H6skw4KHUIsa4q9jkdTDlYColQTUviUhkUpBoRlbuLKS40sfIM9uQGOuwuhw5DQ4DXmxXyPBYD+s8TqYeaENplZY6F5HIoyDRjBzVrSFRL85m8nL7Aga6vKz2uLh+9nIOlWomjohEFo2RaCZM0+TTDbkYBpzfp53V5UgDSbSZvNYhn2kH27BsfwlXP7+MOT8cxpltE469c/UYixPSGAsRaQRqkWgmsnYf5mBJFUO7tiYt0WV1OdKAEm0mczvkc2m/DuwprOCKmUt4b53WmRCRyKAg0UwsWJMDwJUDO1pciTQGlwHP3jCIn1+QQaUvwJ1vrOHed9ZRXKFxEyJiLQWJZsDrD/LB+gM47AaX9e9gdTnSSGw2gzu/15O/TR1OaoKTt1bv4/ynFvPB+v3an0NELKMg0Qx8vvkQxZU+zuvVllZxTqvLkUY2pmca/757HFcN7kReqYfb/76Ga19cxnqPZuqISNNTkGgG/rk21K0xcZC6NVqK1vFOnrh2AH//SSa92yeyatdhrshpy88PteagX7/WItJ09BcnyhVX+vhs4yESY2M4r3dbq8uRJjbyzFQ+vHMMj07qR6o9wLtlcZy3tx1PH06kUotYiUgTUJCIcu+tzcHrD3Jpvw7EOuxWlyMWsNsMbhjehf92zuWW5FICpsHTh5M4b287FpS6CWr4hIg0IgWJKGaaJn9bthuAGzO7WlyNWC3RZnJfmxIWdc7l0vgKDgbs3J2XwsT9aWRr/ISINBIFiSi2fEchWw+VMbBzK/p1Sra6HIkQnR0BZrU7zNvpefR3eVnncTIxJ43ZX2wnqOYJEWlgWtkyir22fBcAk6t2wOwN1hYjEWdYrJd/pufxWkk8fyxM5pGPNvHl1nyeuGYAbZNirS5PRJoJBYkodbC4in9tyCXFFuDS+Eqry5GmcKolsI/DZsCU5HIy3R7u9PXgy635XPTMlzx+TX/G99ZS6iJy+tS1EaX+tmwXgaDJtYkVxOoqyin0dvp57/bR3DSiK4XlXqbOXc2jH2/EHwhaXZqIRDm9BEWhw+VeXl26C6fdxpTkMqvLkSgR67Dz8JV9mT15CImxMby4eAc/eGkFuSXaUVRE6k9BIgq9vGQn5d4A1w3rTIcYvaOUWpo9G2bPZsJX7/Fh2j76Or2s3FXIpX/+lKXb8q2uTkSilIJElCmu8DF36S4cdoNbz+1hdTkSpbo4AryTnseNiWXkB+z8v5dX8Ngnm/D6FUxFpG402DLKvLxkB2UePz/I7EJ6K7fV5UgUi7XBH9OKGRbr5f6Stjz3+Xb+uzmPx67qX/fpxCcbCDpt2ukVKiIRTS0SUWRvYQUvfrEDZ4yN6WqNkAby/cRKPr5rDJndU9h4oIQrZi3hV++s51Cpxk6IyKkpSESRP364EY8/yC3jetCpdZzV5Ugz0jkljjduPoc/TuxLK7eDf6zey+g//5cZ737DltxSq8sTkQimro0o8cWWPD7ZcJCOrdzcOk6tEdLwbDaDGzO7clm/dGb+dyt/X7GHN1aGPnq2TeCCs9oxqEtr+nZMom1iLHabNgUTEQWJqFDu8fPgwmwAHrjsLNxObc4ljSc5zsFvLj2L28f35I2Ve/jnmhw2HSxl66HvphrbbQapCU7aJsbSNtFFal4r2tiDtLEHSLUHaWsP0N3hp509iOKGSPOmIBEFfvveBnYVVHB+n3ZceLZWI5RGcJzBksnALcAtcbDzlzfy1bZ81u8rYvPBUnJLPOSVecgt8VTfO/64p403gvSfvZxBXVoxqEtrBnVpRWqCq9F+DBFpegoSEW7h2hzeydpHuyQXj13dH8PQ+ztpet1T4+meGg98t8tsIGhSWO7lUGkVBW/9k4KAjYKAnfyAjQMBO7t8MWz1xrBsRwHLdhSEv+/MtgmM7ZnGmIxUzuneRi1sIlFOQSKCbTxQwm8WZGMY8NS1A0mJd1pdkkiY3WaQlugiLdEFcZ7j3idgwtYrr2fNniK+3n2YrD2H2XaojG2Hypjz1U6cMTaGd0vhjDgP8R1L6NUuUWFZJMooSESofYcrmDJnJWUeP3efn8HIM1OtLkmkzuwG9H7vTXoDNwAkwn63nS8qXHxZ6WJJZSxLtuWzBPjb+i9pm+hiTM80xvVKY8yZqbRWeBaJeAoSESiv1MOUOSs5VOrhuqGdufN7Z1pdkkiDSY8JcH1SBdcnVRAwYZ3HwfsHqljnbMfaUpP5X+9j/tf7MDDp7/IxbtRZjOuVxoBOrYixa8a6SKRRkIgwO/PLmTJnJXsKKxjfuy1/LFiO8dJyq8uSlq4eW5jXht2AwbE+2tj20bWjneKAwVeVLr6ojOWLChfrPE7W/Wcbf/nPNpJiYxjdM5WxPdMYm5GmlV1FIoSCRARZvqOA217/moJyL5f268AT1w4g5pU1Vpcl0mSS7SaXJFRxSUIVpgnbfDEsrnCxuDKWFVUuPvrmIB99cxCAng4f4+KqGHf19xjWLYVYhwZtilhBQSICeP1Bnvz3Fl78YjumCT8c2Y0HLzsLmxb8kRbMMKCn009Pp5+ftCqnMmiwosrJFxWxLK50sdXnYGuxg7++vJJYh40+HZLIaJtIRvtEerVLpFtqHGmJLlwxChgijUlBwkKmafJx9kEe+2QTuwoqSHTF8IeJfblyYEerSxOJOG6byblxHs6tniGS4w8N2lzcrg9fbc9nzZ4i1uwpOub7kt0O0hJdpCY4aR3nJNntIDnOQSu3k1ZxDpLdDlpV35YS79SqnSJ1pCBhgUpvgPfW5fDq0t18e6AEgLHuKh5JLaLTyt2w0uICRaJAx5gANyRVcMPkIQSDJjlFlWw9VMrmg2VsPljCvsOV5OXkcagqyLZKH9sO1e68dkzaxwToYA+Q3rs7HVrF0rGVm46t3HRqHUfH1m4SXPrTKVLD0t8Gn8/Ho48+yvvvv49hGFx++eXMmDGDmJjm90t6sLiKFTsL+PTbXD7fdIhybwCAfk4v96aUMOYE8/BF5BRmz8YGdK7+GF9zux3oEvq0PGiQF7BRFLBRHLRRFLRRfMTnRQGD4qCN/ICdA347+/12cvwxsG7/cR+yVZyDTq3ddGoVChadWodCRqfWbjq2dpMU62j0H1skUlj6iv3888+TlZXFhx9+CMDNN9/MCy+8wO23325lWfXm8QfIL/OSc7iSHXllbM8rY0deOZtzS9l3uDJ8v1iHjSsGpHPTiK4M+dc7aP0dkcYVbzOJtwXAEajV/b0m5FYHigN+OznVwWKf384+v52cSpPsCh/ZOSXH/f44I0hq6wRSE5ykJrhITXSRmuCiTbyTWIeNWIcdV4wNV4ydoGniCwTxBkx8/iAef5AKr5/KpSuoMA0qgwaVpkHFEf8GMTAAm2FiI7SNs9MwcRomLgOcfXrhirHhrPmw23HG2HDF2HDYDQJLviIIBE0IYhAktHiYSfXXQ4YQNAHTxAT2HyhlUd4mTMBcsxaz+vFiDZNYW+hfd/XX7ksvIik2hqRYB4mxDpLcMbgddi001oxZGiTmz5/PjBkzaNu2LQC33HILjz32WJMEieIKH8t2FOAPBgkETYKmSSAIwaBJwDSPuC304QuYlHv8lHv9VHgClHn9VHj8lHsCFJR7OFhUQfnbnxz3sQxMejv9DHF5GBvnYazbg7t0L3y6Cu1oJBJ5nAZ0dgTofILgYZqQH7CFQsWRAcMX+vxQwM6ewgr2FFacRhVJ9f/WVXtPcYdWJz/82dZjb9u4vfqTxJN/7yurjrnJbjNC4cLtIDEcMr4LG4mxMThjbMTYDGLsNf8aoX9tNmLsBnabUR2fGlbQDP2t9weq/96bJv6gSSAQJGBCIBjEHzTx+UOBLxT6gvgDZvhzX8DEt3U7fsBrGvhMA58JPgwCJjjat8Vpt4XDnDPGhsNuC98WCnvf3R76MI762m4DAwPDAMMIPROGEbpt574qDrtzGdGjDfEWdLtZFiSKi4s5ePAgffr0Cd/Wp08f9u/fT2lpKYmJof+spmkC4PV6G/TxH//kWz7Jzm2w87WKtXFmWixt4p2kJbrokhJH1zZxdF31JZ1iAsTZzCPu7UAdGU0rEBeHx6lVEiNVNF6fJOAs4CwCwLGBw3ftdRyu8HK43EdhuZfCci9FlV68fhOPP4C3uvXBblS/aNqN0IuJ3Uas045rTdZ37/INE7fNxGUEibWF/nCHWg9qWhVCL2Bek9CL2Lnj8fqDR7zQBfH5q18Agya2tWuxEXohshHqBTKOaN0wMjOPesHat28fXTp3Ct22dBkAfsBjGlQFQ49bZRpUmTYqTYPyoEFZ0EZZ0KDcNCgN2igP+qmoDFJR6aHh/vJGkoTQPwbHvkHML238h1+3hh9kdmb6uQ27gGHNa2/Na/HxGObJjjaiAwcOcO6557Js2TJSUlIAKCwsZMSIESxevJj27dsDUFpaypYtW6woUURERICMjIzwG/z/ZVmLRFxcHABlZWXhIFFaGkpt8fHfbUkcHx9PRkYGDodDfWwiIiJNyDRNfD7fUa/L/8uyIJGcnEz79u3ZuHEjXbqEhlZv3LiRDh06HJV6bDbbCVOQiIiINK7Y2NiTHrd0B5xJkybxwgsvkJeXR15eHi+++CJXX321lSWJiIhIHVg6a2P69OkUFRVxySWXAHDFFVdwyy23WFmSiIiI1IGlLRIOh4Pf/va3rFq1ilWrVvHAAw9E3WJUPp+Phx9+mGHDhjF8+HB+//vf4/f7rS4rKnm9Xu6//37Gjx/PoEGDuOiii3jnnXfCx8vKyvjFL37B4MGDGTlyJLNmzTrq+60+3pJUVVVxwQUXMHTo0PBtVj//uj4hixYt4sorr2TgwIGMHj2aN954A7D++df1gdzcXKZPn05mZiaZmZncddddFBYWAqd+LbH6+EmZclqeeeYZ84orrjBzc3PN3Nxc84orrjCfffZZq8uKSuXl5ebTTz9t7t692wwGg+aaNWvMoUOHml9++aVpmqZ57733mlOnTjWLi4vNHTt2mOPGjTMXLFgQ/n6rj7ckf/rTn8ybbrrJHDJkSPg2q59/XR/TXLx4sTlmzBhz+fLlpt/vN4uKisxt27aZpmn986/rY5q33nqreeutt5plZWVmaWmp+dOf/tS8++67TdM89WuJ1cdPRkHiNI0dO9b8+OOPw19/9NFH5rnnnmthRc3LbbfdZj799NNmRUWFefbZZ5vr168PH3vppZfMG2+80TRN0/LjLck333xjXnbZZeaXX34ZDhJWP/+6PiGTJk0y33zzzWNut/r51/UJueyyy8z33nsv/PXChQvNSy+91DTNU7+WWH38ZCzt2oh2p1pUS06Px+Nh/fr19OrVi507d+Lz+Y55rjdv3gxg+fGWwu/388ADD/Dggw/icHy3n4TVz7+uD1RUVLBhwwZyc3O58MILGTVqFHfeeSeHDh2y/PnX9Qn50Y9+xCeffEJpaSklJSV8+OGHnHfeead8LbH6+KkoSJyGiorQ8rdHTk9NSgota1teXm5JTc2FaZr85je/oWvXrkyYMIGKigri4uKOGkOTmJgYfp6tPt5SvPzyy/Tp04dhw4YddbvVz7+uD5SUlGCaJp999hlz5szh008/xel0cs8991j+/Ov6hAwePJiCgoLwOITi4mJ++tOfnvK1xOrjp6IgcRqOXFSrxvEW1ZK6MU2T3/3ud+zcuZPnnnsOm81GXFwclZWVRw3+KSsrCz/PVh9vCXbv3s2bb77Jvffee8wxq59/XZ/v/h5NnjyZjh07Eh8fz5133smKFSswDEPXx2LBYJCpU6cyePBg1qxZw5o1axg8eDBTp0495WuJ1cdPRUHiNBy5qFaN4y2qJbVnmiYPPfQQ69evZ86cOeHnsXv37sTExLBp06bwfTdu3EhGRkZEHG8JsrKyyM/P58ILLyQzM5Pp06dTVlZGZmYmZWVluj4WS0pKIj09/bjHevXqpetjsaKiInJycrjppptwu9243W4mT57MunXrCAQCJ30tOdVrTWMfPxUFidOkRbUa1sMPP8zXX3/NnDlzSE5ODt/udru55JJLeOaZZygtLWXXrl3MmzePa665JiKOtwQXX3wx//73v1m4cCELFy7kD3/4A/Hx8SxcuJCBAwfq+kSAa6+9lnnz5pGbm0tVVRWzZs1ixIgRJCQk6PpYLCUlha5du/L666/j8XjweDy8/vrrtG/fnpSUlFO+llh9/KRqNSRTTsjr9Zq/+93vzKFDh5pDhw41H374YdPn81ldVlTat2+fmZGRYfbt29ccOHBg+OOBBx4wTdM0S0tLzbvvvtscOHCgec455xwzNcnq4y3N8uXLj5r+afXzr+tjmn6/33z00UfN4cOHm8OHDzfvuOMO89ChQ6ZpWv/86/qY5tatW82pU6eaw4cPN4cOHWpOnjzZ3LBhg2map34tsfr4yVi2+6eIiIhEP3VtiIiISL0pSIiIiEi9KUiIiIhIvSlIiIiISL0pSIiIiEi9KUiIiIhIvSlIiIiISL3FnPouIhItvv76a2bNmsXmzZspLi4mOTmZTp060bdvX2bMmIHdbq/1ucaPH09OTg7Dhw/ntddeq1Md7777LjNmzDjqtpiYGNq1a8d5553HnXfeedTKpSeyb98+vve97wFw++23c8cdd9SpDhFpfGqREGkmvvrqK2688UaWLFlCXl4eXq+XvLw81qxZw2uvvXbUhklW8Pv95OTkMG/ePG688UY8Ho+l9YhIw1CLhEgzMWfOHILBIB07duSFF16gW7duHDp0iG+++YaFCxdiGIYlddW0JOTl5fGzn/2M1atXs3XrVt5///2TruVfVVVFp06d2Lx5cxNWKyJ1pRYJkWZi3759QGinxYyMDJxOJ506deLiiy/mhRdewOl0AjB79myuueYazjnnHM4++2yGDRvGj370I5YvX16rx/nyyy+ZMmUKQ4YMoV+/flx++eW8/vrrnGq1/bS0NH74wx+Gv87Ozgbg2WefpVevXvTq1Ytly5YxdepUBgwYwBNPPMG+ffvCx5599tnw95qmyT/+8Q+uvvpqBg0axMCBA7n88st59913w/fxer3MnDmTiy++mH79+jF06FCmTZsWflwRaRhqkRBpJtq2bcuuXbtYsmQJV111FWPGjGHo0KEMHTqU2NjY8P0+++wz1q9fH/66pKSEpUuXsmrVKt555x169+59wsd4++23uf/++4+6bcuWLTz88MNs376dBx988KQ1Hhk2jtdCctddd1FcXHzKn/XXv/71UaGhpo6VK1cyadIk/H4/P/nJT1ixYkX4uNfrZfHixSxbtoy5c+cyZMiQUz6OiJyaWiREmokbbrgh/Hl2djbPP/88P/7xjxk1ahQvvvhi+Nj06dP56KOPyMrKIjs7m/nz5+N2u/H5fLzzzjsnPH95eTl/+tOfAJgwYQJLlixhzZo1TJ06FYC///3vbN++/YTfn5eXx9y5c8Nfn3XWWcfcp02bNrz33nt8/fXX3HTTTcc9z+rVq8Mholu3brz99tusWbOGt956ixEjRgDwwQcfhEPEn//8Z9avX8+iRYvo0aMHXq+XP//5zyesU0TqRi0SIs3EJZdcgsPh4Pnnn2fDhg3h28vKynjyySfp3Lkzl1xyCYmJiTzyyCNkZ2dTUlJCMBgM33fnzp0nPP+aNWsoKysD4NNPP+XTTz896rhpmqxYsYIePXocdfvMmTOZOXPmUbd1796dyy+//JjHuPPOO+nVqxcA8fHx4e6aI33xxRfhz++77z769+8PwIABAxgwYAAQ6n6p8atf/Ypf/epXR51j/fr1VFZW4na7T/jzikjtKEiINCMXXHABF1xwAQcOHGDZsmX84x//YO3atQAsWrSI/v37c/PNN1NeXn7c7/d6vSc8d2Fh4Skf/2TdEna7nbS0NM4991zuuuuuo7pbamRkZJzyMY6so3v37vWq1TRNSkpKFCREGoCChEgzUVZWRkJCAgAdOnRg0qRJTJgwITwWoKioiCVLloRDxIMPPsg111yD0+kkMzOToqKik56/devW4c9//etfM2XKlKOOm6Z53HEPdVn/weVynfI+KSkp4c937txJt27dTlirzWZj+fLlx6xZcaJaRaTuNEZCpJm49dZbeeCBB1ixYgVlZWVUVlby/vvvh493794dn88X/jo+Ph6/389LL710yhABMGjQIOLj4wF4+eWXycrKwuv1kpuby4IFC5g4cWKD/0zHM3bs2PDnjz32GN988w2VlZVkZ2ezcOFCAMaMGQNAMBjkt7/9Lbm5uXi9XjZt2sRTTz3FI4880iS1irQEapEQaSY8Hg9vvfUWb7311jHH4uLi+MEPfgCAw+HA5/OFxw60atWKpKQkSkpKTnr+hIQE7r333vALc835mtrQoUOZNGkS7777Ljt27DhqLYqJEydy5ZVXctlllzF//nxWrVrFxx9/zMcff3zUOZoq9Ii0BGqREGkmfvazn3H99deTkZFB69atiYmJISUlhfPPP5958+ZxxhlncMYZZ/DUU09xxhln4HK56N+/P3/9619JTEys1WNcf/31vPTSS4wYMYLExMTwWhUTJkzg8ccfb+Sf8DuPPPIIDz/8MP369cPtdhMbG0tGRgbDhw8HQmHp5Zdf5o477qBHjx44nU4SExPJyMhgypQpR61nISKnxzBPtYqMiIiIyAmoRUJERETqTUFCRERE6k1BQkREROpNQUJERETqTUFCRERE6k1BQkREROpNQUJERETqTUFCRERE6k1BQkREROpNQUJERETqTUFCRERE6u3/A+xjwkhxnTpTAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 540x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(dpi=90)\n",
    "ax = sns.distplot(y, kde=True, hist=False)\n",
    "ax = sns.distplot(X[\"Count\"], color='r', ax=ax, hist=True, kde=False, norm_hist=True)\n",
    "ax.set_xlabel(\"SalePrice\");"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2a2d44de",
   "metadata": {
    "papermill": {
     "duration": 0.024084,
     "end_time": "2022-04-10T23:34:08.466450",
     "exception": false,
     "start_time": "2022-04-10T23:34:08.442366",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "And the distributions are almost exactly the same, too.\n",
    "\n",
    "# 3) Overfitting with Target Encoders\n",
    "\n",
    "Based on your understanding of how mean-encoding works, can you explain how XGBoost was able to get an almost a perfect fit after mean-encoding the count feature?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "1c89b30e",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-10T23:34:08.516639Z",
     "iopub.status.busy": "2022-04-10T23:34:08.516111Z",
     "iopub.status.idle": "2022-04-10T23:34:08.524765Z",
     "shell.execute_reply": "2022-04-10T23:34:08.524313Z",
     "shell.execute_reply.started": "2022-04-10T23:30:56.435692Z"
    },
    "lines_to_next_cell": 0,
    "papermill": {
     "duration": 0.03511,
     "end_time": "2022-04-10T23:34:08.524871",
     "exception": false,
     "start_time": "2022-04-10T23:34:08.489761",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "application/javascript": [
       "parent.postMessage({\"jupyterEvent\": \"custom.exercise_interaction\", \"data\": {\"outcomeType\": 1, \"valueTowardsCompletion\": 0.3333333333333333, \"interactionType\": 1, \"questionType\": 4, \"questionId\": \"3_Q3\", \"learnToolsVersion\": \"0.3.4\", \"failureMessage\": \"\", \"exceptionClass\": \"\", \"trace\": \"\"}}, \"*\")"
      ],
      "text/plain": [
       "<IPython.core.display.Javascript object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "<span style=\"color:#33cc33\">Correct:</span> \n",
       "\n",
       "\n",
       "Since `Count` never has any duplicate values, the mean-encoded `Count` is essentially an exact copy of the target. In other words, mean-encoding turned a completely meaningless feature into a perfect feature.\n",
       "\n",
       "Now, the only reason this worked is because we trained XGBoost on the same set we used to train the encoder. If we had used a hold-out set instead, none of this \"fake\" encoding would have transferred to the training data.\n",
       "\n",
       "The lesson is that when using a target encoder it's very important to use separate data sets for training the encoder and training the model. Otherwise the results can be very disappointing!\n"
      ],
      "text/plain": [
       "Correct: \n",
       "\n",
       "\n",
       "Since `Count` never has any duplicate values, the mean-encoded `Count` is essentially an exact copy of the target. In other words, mean-encoding turned a completely meaningless feature into a perfect feature.\n",
       "\n",
       "Now, the only reason this worked is because we trained XGBoost on the same set we used to train the encoder. If we had used a hold-out set instead, none of this \"fake\" encoding would have transferred to the training data.\n",
       "\n",
       "The lesson is that when using a target encoder it's very important to use separate data sets for training the encoder and training the model. Otherwise the results can be very disappointing!"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# View the solution (Run this cell to receive credit!)\n",
    "q_3.check()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "fed7e135",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-10T23:34:08.577038Z",
     "iopub.status.busy": "2022-04-10T23:34:08.576570Z",
     "iopub.status.idle": "2022-04-10T23:34:08.579520Z",
     "shell.execute_reply": "2022-04-10T23:34:08.579136Z",
     "shell.execute_reply.started": "2022-04-10T23:32:04.003470Z"
    },
    "papermill": {
     "duration": 0.03057,
     "end_time": "2022-04-10T23:34:08.579624",
     "exception": false,
     "start_time": "2022-04-10T23:34:08.549054",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Uncomment this if you'd like a hint before seeing the answer\n",
    "#q_3.hint()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6a86c9e3",
   "metadata": {
    "papermill": {
     "duration": 0.023812,
     "end_time": "2022-04-10T23:34:08.627335",
     "exception": false,
     "start_time": "2022-04-10T23:34:08.603523",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# The End #\n",
    "\n",
    "That's it for *Feature Engineering*! We hope you enjoyed your time with us.\n",
    "\n",
    "Now, are you ready to try out your new skills? Now would be a great time to join our [Housing Prices](https://www.kaggle.com/c/house-prices-advanced-regression-techniques) Getting Started competition. We've even prepared a [Bonus Lesson]() that collects all the work we've done together into a starter notebook.\n",
    "\n",
    "# References #\n",
    "Here are some great resources you might like to consult for more information. They all played a part in shaping this course:\n",
    "- *The Art of Feature Engineering*, a book by Pablo Duboue.\n",
    "- *An Empirical Analysis of Feature Engineering for Predictive Modeling*, an article by Jeff Heaton.\n",
    "- *Feature Engineering for Machine Learning*, a book by Alice Zheng and Amanda Casari. The tutorial on clustering was inspired by this excellent book.\n",
    "- *Feature Engineering and Selection*, a book by Max Kuhn and Kjell Johnson."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f12bbad9",
   "metadata": {
    "papermill": {
     "duration": 0.023444,
     "end_time": "2022-04-10T23:34:08.674862",
     "exception": false,
     "start_time": "2022-04-10T23:34:08.651418",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "---\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "*Have questions or comments? Visit the [course discussion forum](https://www.kaggle.com/learn/feature-engineering/discussion) to chat with other learners.*"
   ]
  }
 ],
 "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": "3.7.12"
  },
  "papermill": {
   "default_parameters": {},
   "duration": 27.736258,
   "end_time": "2022-04-10T23:34:09.611043",
   "environment_variables": {},
   "exception": null,
   "input_path": "__notebook__.ipynb",
   "output_path": "__notebook__.ipynb",
   "parameters": {},
   "start_time": "2022-04-10T23:33:41.874785",
   "version": "2.3.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}