intermediate-machine-learning-notes.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "4b04ffba",
   "metadata": {
    "papermill": {
     "duration": 0.009138,
     "end_time": "2022-10-10T14:20:12.174804",
     "exception": false,
     "start_time": "2022-10-10T14:20:12.165666",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Intermediate Machine Learning\n",
    "\n",
    "In this module, the focus will be on the following:\n",
    "\n",
    "- design pipelines to improve the quality of your machine learning code,\n",
    "- use advanced techniques for model validation (cross-validation),\n",
    "- build state-of-the-art models that are widely used to win Kaggle competitions (XGBoost), and\n",
    "- avoid common and important data science mistakes (leakage).\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2f572b5d",
   "metadata": {
    "papermill": {
     "duration": 0.007559,
     "end_time": "2022-10-10T14:20:12.190061",
     "exception": false,
     "start_time": "2022-10-10T14:20:12.182502",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Pipelines\n",
    "\n",
    "Pipelines are a simple way to keep your data preprocessing and modeling code organized. Specifically, a pipeline bundles preprocessing and modeling steps so you can use the whole bundle as if it were a single step.\n",
    "\n",
    "Main benefits of using pipeline are :\n",
    "- Cleaner Code: Accounting for data at each step of preprocessing can get messy. With a pipeline, you won't need to manually keep track of your training and validation data at each step.\n",
    "- Fewer Bugs: There are fewer opportunities to misapply a step or forget a preprocessing step.\n",
    "- Easier to Productionize: It can be surprisingly hard to transition a model from a prototype to something deployable at scale. We won't go into the many related concerns here, but pipelines can help.\n",
    "\n",
    "The full pipeline can be constructed in three steps:\n",
    "### **Step 1**: **Define Preprocessing Steps**:\n",
    "    - imputes missing values in numerical data, and\n",
    "    - imputes missing values and applies a one-hot encoding to categorical data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "c64a6bbf",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-10-10T14:20:12.207009Z",
     "iopub.status.busy": "2022-10-10T14:20:12.206541Z",
     "iopub.status.idle": "2022-10-10T14:20:13.528244Z",
     "shell.execute_reply": "2022-10-10T14:20:13.526808Z"
    },
    "papermill": {
     "duration": 1.333927,
     "end_time": "2022-10-10T14:20:13.531508",
     "exception": false,
     "start_time": "2022-10-10T14:20:12.197581",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "# Read the data\n",
    "data = pd.read_csv('../input/melbourne-housing-snapshot/melb_data.csv')\n",
    "\n",
    "# Separate target from predictors\n",
    "y = data.Price\n",
    "X = data.drop(['Price'], axis=1)\n",
    "\n",
    "# Divide data into training and validation subsets\n",
    "X_train_full, X_valid_full, y_train, y_valid = train_test_split(X, y, train_size=0.8, test_size=0.2,\n",
    "                                                                random_state=0)\n",
    "\n",
    "# \"Cardinality\" means the number of unique values in a column\n",
    "# Select categorical columns with relatively low cardinality (convenient but arbitrary)\n",
    "categorical_cols = [cname for cname in X_train_full.columns if X_train_full[cname].nunique() < 10 and \n",
    "                        X_train_full[cname].dtype == \"object\"]\n",
    "\n",
    "# Select numerical columns\n",
    "numerical_cols = [cname for cname in X_train_full.columns if X_train_full[cname].dtype in ['int64', 'float64']]\n",
    "\n",
    "# Keep selected columns only\n",
    "my_cols = categorical_cols + numerical_cols\n",
    "X_train = X_train_full[my_cols].copy()\n",
    "X_valid = X_valid_full[my_cols].copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "2f619363",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-10-10T14:20:13.548798Z",
     "iopub.status.busy": "2022-10-10T14:20:13.547626Z",
     "iopub.status.idle": "2022-10-10T14:20:13.707858Z",
     "shell.execute_reply": "2022-10-10T14:20:13.706651Z"
    },
    "papermill": {
     "duration": 0.172122,
     "end_time": "2022-10-10T14:20:13.711034",
     "exception": false,
     "start_time": "2022-10-10T14:20:13.538912",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "from sklearn.compose import ColumnTransformer\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.impute import SimpleImputer\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "\n",
    "# Preprocessing for numerical data\n",
    "numerical_transformer = SimpleImputer(strategy='constant')\n",
    "\n",
    "# Preprocessing for categorical data\n",
    "categorical_transformer = Pipeline(steps=[\n",
    "    ('imputer', SimpleImputer(strategy='most_frequent')),\n",
    "    ('onehot', OneHotEncoder(handle_unknown='ignore'))\n",
    "])\n",
    "\n",
    "# Bundle preprocessing for numerical and categorical data\n",
    "preprocessor = ColumnTransformer(\n",
    "    transformers=[\n",
    "        ('num', numerical_transformer, numerical_cols),\n",
    "        ('cat', categorical_transformer, categorical_cols)\n",
    "    ])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a6c75a10",
   "metadata": {
    "papermill": {
     "duration": 0.007062,
     "end_time": "2022-10-10T14:20:13.725707",
     "exception": false,
     "start_time": "2022-10-10T14:20:13.718645",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "### Step 2: Define the Model:\n",
    "Next, we define a `random forest model` with the familiar `RandomForestRegressor` class."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "484d6ae5",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-10-10T14:20:13.742690Z",
     "iopub.status.busy": "2022-10-10T14:20:13.742271Z",
     "iopub.status.idle": "2022-10-10T14:20:13.846103Z",
     "shell.execute_reply": "2022-10-10T14:20:13.845030Z"
    },
    "papermill": {
     "duration": 0.115668,
     "end_time": "2022-10-10T14:20:13.848924",
     "exception": false,
     "start_time": "2022-10-10T14:20:13.733256",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestRegressor\n",
    "\n",
    "model = RandomForestRegressor(n_estimators=100, random_state=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cbe0c735",
   "metadata": {
    "papermill": {
     "duration": 0.007106,
     "end_time": "2022-10-10T14:20:13.863528",
     "exception": false,
     "start_time": "2022-10-10T14:20:13.856422",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "### Step 3: Create and Evaluate the Pipeline\n",
    "- Finally, we use the Pipeline class to define a pipeline that bundles the preprocessing and modeling steps. There are a few important things to notice:\n",
    "- With the pipeline, we preprocess the training data and fit the model in a single line of code. (In contrast, without a pipeline, we have to do imputation, one-hot encoding, and model training in separate steps. This becomes especially messy if we have to deal with both numerical and categorical variables!)\n",
    "- With the pipeline, we supply the unprocessed features in X_valid to the predict() command, and the pipeline automatically preprocesses the features before generating predictions. (However, without a pipeline, we have to remember to preprocess the validation data before making predictions.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "3a0a1afe",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-10-10T14:20:13.880635Z",
     "iopub.status.busy": "2022-10-10T14:20:13.879941Z",
     "iopub.status.idle": "2022-10-10T14:20:21.217337Z",
     "shell.execute_reply": "2022-10-10T14:20:21.216077Z"
    },
    "papermill": {
     "duration": 7.349163,
     "end_time": "2022-10-10T14:20:21.220136",
     "exception": false,
     "start_time": "2022-10-10T14:20:13.870973",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MAE: 160679.18917034855\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import mean_absolute_error\n",
    "\n",
    "# Bundle preprocessing and modeling code in a pipeline\n",
    "my_pipeline = Pipeline(steps=[('preprocessor', preprocessor),\n",
    "                             ('model', model)\n",
    "                            ])\n",
    "\n",
    "# Preprocessing of training data, fit model \n",
    "my_pipeline.fit(X_train, y_train)\n",
    "\n",
    "# Preprocessing of validation data, get predictions\n",
    "preds = my_pipeline.predict(X_valid)\n",
    "\n",
    "# Evaluate the model\n",
    "score = mean_absolute_error(y_valid, preds)\n",
    "print('MAE:', score)"
   ]
  },
  {
   "attachments": {
    "5bbf656d-2ab7-469b-b219-bb5281a2591a.png": {
     "image/png": "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"
    }
   },
   "cell_type": "markdown",
   "id": "b6aa5e06",
   "metadata": {
    "papermill": {
     "duration": 0.008239,
     "end_time": "2022-10-10T14:20:21.235902",
     "exception": false,
     "start_time": "2022-10-10T14:20:21.227663",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Cross Validation\n",
    "\n",
    "In cross-validation, we run our modeling process on different subsets of the data to get multiple measures of model quality.\n",
    "![image.png](attachment:5bbf656d-2ab7-469b-b219-bb5281a2591a.png)\n",
    "\n",
    "**Cross-validation** gives a more accurate measure of model quality, which is especially important if you are making a lot of modeling decisions. However, it can take longer to run, because it estimates multiple models (one for each fold).\n",
    "\n",
    "- For small datasets, where extra computational burden isn't a big deal, you should run cross-validation.\n",
    "- For larger datasets, a single validation set is sufficient. Your code will run faster, and you may have enough data that there's little need to re-use some of it for holdout."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "2f9ef1fe",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-10-10T14:20:21.253798Z",
     "iopub.status.busy": "2022-10-10T14:20:21.253408Z",
     "iopub.status.idle": "2022-10-10T14:20:28.381425Z",
     "shell.execute_reply": "2022-10-10T14:20:28.380022Z"
    },
    "papermill": {
     "duration": 7.140758,
     "end_time": "2022-10-10T14:20:28.384740",
     "exception": false,
     "start_time": "2022-10-10T14:20:21.243982",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MAE scores:\n",
      " [301628.7893587  303164.4782723  287298.331666   236061.84754543\n",
      " 260383.45111427]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestRegressor\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.impute import SimpleImputer\n",
    "from sklearn.model_selection import cross_val_score\n",
    "import pandas as pd\n",
    "\n",
    "# Read the data\n",
    "data = pd.read_csv('../input/melbourne-housing-snapshot/melb_data.csv')\n",
    "\n",
    "# Select subset of predictors\n",
    "cols_to_use = ['Rooms', 'Distance', 'Landsize', 'BuildingArea', 'YearBuilt']\n",
    "X = data[cols_to_use]\n",
    "\n",
    "# Select target\n",
    "y = data.Price\n",
    "my_pipeline = Pipeline(steps=[('preprocessor', SimpleImputer()),\n",
    "                              ('model', RandomForestRegressor(n_estimators=50,\n",
    "                                                              random_state=0))])\n",
    "\n",
    "\n",
    "\n",
    "# Multiply by -1 since sklearn calculates *negative* MAE\n",
    "scores = -1 * cross_val_score(my_pipeline, X, y,\n",
    "                              cv=5,\n",
    "                              scoring='neg_mean_absolute_error')\n",
    "\n",
    "print(\"MAE scores:\\n\", scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "05431df5",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-10-10T14:20:28.404103Z",
     "iopub.status.busy": "2022-10-10T14:20:28.403704Z",
     "iopub.status.idle": "2022-10-10T14:20:28.409566Z",
     "shell.execute_reply": "2022-10-10T14:20:28.408385Z"
    },
    "papermill": {
     "duration": 0.017449,
     "end_time": "2022-10-10T14:20:28.412108",
     "exception": false,
     "start_time": "2022-10-10T14:20:28.394659",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average MAE score (across experiments):\n",
      "277707.3795913405\n"
     ]
    }
   ],
   "source": [
    "print(\"Average MAE score (across experiments):\")\n",
    "print(scores.mean())"
   ]
  },
  {
   "attachments": {
    "900ad00d-9eb4-4e03-af65-9b1c8eb5ed1a.png": {
     "image/png": "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"
    }
   },
   "cell_type": "markdown",
   "id": "93b25010",
   "metadata": {
    "papermill": {
     "duration": 0.008301,
     "end_time": "2022-10-10T14:20:28.428130",
     "exception": false,
     "start_time": "2022-10-10T14:20:28.419829",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Using cross-validation yields a much better measure of model quality, with the added benefit of cleaning up our code.\n",
    "\n",
    "## XGBoost\n",
    "\n",
    "It is considered as the most accurate modeling technique for structured data.\n",
    "\n",
    "Gradient boosting is a method that goes through cycles to iteratively add models into an ensemble.\n",
    "\n",
    "![image.png](attachment:900ad00d-9eb4-4e03-af65-9b1c8eb5ed1a.png)\n",
    "\n",
    "**XGBoost** stands for extreme gradient boosting, which is an implementation of gradient boosting with several additional features focused on performance and speed. (Scikit-learn has another version of gradient boosting, but XGBoost has some technical advantages.)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "a419b283",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-10-10T14:20:28.444539Z",
     "iopub.status.busy": "2022-10-10T14:20:28.444117Z",
     "iopub.status.idle": "2022-10-10T14:20:29.273746Z",
     "shell.execute_reply": "2022-10-10T14:20:29.272870Z"
    },
    "papermill": {
     "duration": 0.841067,
     "end_time": "2022-10-10T14:20:29.276653",
     "exception": false,
     "start_time": "2022-10-10T14:20:28.435586",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mean Absolute Error: 245378.67665224595\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "from sklearn.model_selection import train_test_split\n",
    "from xgboost import XGBRegressor\n",
    "from sklearn.metrics import mean_absolute_error\n",
    "\n",
    "# Read the data\n",
    "data = pd.read_csv('../input/melbourne-housing-snapshot/melb_data.csv')\n",
    "\n",
    "# Select subset of predictors\n",
    "cols_to_use = ['Rooms', 'Distance', 'Landsize', 'BuildingArea', 'YearBuilt']\n",
    "X = data[cols_to_use]\n",
    "\n",
    "# Select target\n",
    "y = data.Price\n",
    "\n",
    "# Separate data into training and validation sets\n",
    "X_train, X_valid, y_train, y_valid = train_test_split(X, y)\n",
    "\n",
    "\n",
    "my_model = XGBRegressor()\n",
    "my_model.fit(X_train, y_train)\n",
    "\n",
    "\n",
    "predictions = my_model.predict(X_valid)\n",
    "print(\"Mean Absolute Error: \" + str(mean_absolute_error(predictions, y_valid)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "214b0ae1",
   "metadata": {
    "papermill": {
     "duration": 0.00746,
     "end_time": "2022-10-10T14:20:29.294047",
     "exception": false,
     "start_time": "2022-10-10T14:20:29.286587",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "### Parameter Tuning\n",
    "\n",
    "XGBoost have a few parameters that can be effectivily tuned to increase the accuracy and efficiency.\n",
    "\n",
    "The main parameters are :\n",
    "- n_estimators\n",
    "- early_stopping_rounds\n",
    "- learning_rate\n",
    "- n_jobs\n",
    "\n",
    "#### n_estimators\n",
    "\n",
    "**n_estimators** specifies how many times to go through the modeling cycle described above. It is equal to the number of models that we include in the ensemble.\n",
    "- Too low a value causes underfitting, which leads to inaccurate predictions on both training data and test data.\n",
    "- Too high a value causes overfitting, which causes accurate predictions on training data, but inaccurate predictions on test data (which is what we care about)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "0b284a63",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-10-10T14:20:29.311124Z",
     "iopub.status.busy": "2022-10-10T14:20:29.310733Z",
     "iopub.status.idle": "2022-10-10T14:20:32.431311Z",
     "shell.execute_reply": "2022-10-10T14:20:32.429789Z"
    },
    "papermill": {
     "duration": 3.133539,
     "end_time": "2022-10-10T14:20:32.435395",
     "exception": false,
     "start_time": "2022-10-10T14:20:29.301856",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "XGBRegressor(base_score=0.5, booster='gbtree', callbacks=None,\n",
       "             colsample_bylevel=1, colsample_bynode=1, colsample_bytree=1,\n",
       "             early_stopping_rounds=None, enable_categorical=False,\n",
       "             eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',\n",
       "             importance_type=None, interaction_constraints='',\n",
       "             learning_rate=0.300000012, max_bin=256, max_cat_to_onehot=4,\n",
       "             max_delta_step=0, max_depth=6, max_leaves=0, min_child_weight=1,\n",
       "             missing=nan, monotone_constraints='()', n_estimators=500, n_jobs=0,\n",
       "             num_parallel_tree=1, predictor='auto', random_state=0, reg_alpha=0,\n",
       "             reg_lambda=1, ...)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "my_model = XGBRegressor(n_estimators=500)\n",
    "my_model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "539dffb5",
   "metadata": {
    "papermill": {
     "duration": 0.008203,
     "end_time": "2022-10-10T14:20:32.451769",
     "exception": false,
     "start_time": "2022-10-10T14:20:32.443566",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "#### early_stopping_rounds\n",
    "\n",
    "**early_stopping_rounds** offers a way to automatically find the ideal value for n_estimators. Early stopping causes the model to stop iterating when the validation score stops improving, even if we aren't at the hard stop for n_estimators. It's smart to set a high value for **n_estimators** and then use **early_stopping_rounds** to find the optimal time to stop iterating."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "3ac1e750",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-10-10T14:20:32.469649Z",
     "iopub.status.busy": "2022-10-10T14:20:32.469237Z",
     "iopub.status.idle": "2022-10-10T14:20:32.863260Z",
     "shell.execute_reply": "2022-10-10T14:20:32.862311Z"
    },
    "papermill": {
     "duration": 0.406419,
     "end_time": "2022-10-10T14:20:32.866105",
     "exception": false,
     "start_time": "2022-10-10T14:20:32.459686",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.7/site-packages/xgboost/sklearn.py:797: UserWarning: `early_stopping_rounds` in `fit` method is deprecated for better compatibility with scikit-learn, use `early_stopping_rounds` in constructor or`set_params` instead.\n",
      "  UserWarning,\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "XGBRegressor(base_score=0.5, booster='gbtree', callbacks=None,\n",
       "             colsample_bylevel=1, colsample_bynode=1, colsample_bytree=1,\n",
       "             early_stopping_rounds=None, enable_categorical=False,\n",
       "             eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',\n",
       "             importance_type=None, interaction_constraints='',\n",
       "             learning_rate=0.300000012, max_bin=256, max_cat_to_onehot=4,\n",
       "             max_delta_step=0, max_depth=6, max_leaves=0, min_child_weight=1,\n",
       "             missing=nan, monotone_constraints='()', n_estimators=500, n_jobs=0,\n",
       "             num_parallel_tree=1, predictor='auto', random_state=0, reg_alpha=0,\n",
       "             reg_lambda=1, ...)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "my_model = XGBRegressor(n_estimators=500)\n",
    "my_model.fit(X_train, y_train, \n",
    "             early_stopping_rounds=5, \n",
    "             eval_set=[(X_valid, y_valid)],\n",
    "             verbose=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7999ea70",
   "metadata": {
    "papermill": {
     "duration": 0.007433,
     "end_time": "2022-10-10T14:20:32.881639",
     "exception": false,
     "start_time": "2022-10-10T14:20:32.874206",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "#### learning_rate\n",
    "\n",
    "Instead of getting predictions by simply adding up the predictions from each component model, we can multiply the predictions from each model by a small number (known as the learning rate) before adding them in.\n",
    "\n",
    "In general, a small learning rate and large number of estimators will yield more accurate XGBoost models, though it will also take the model longer to train since it does more iterations through the cycle. As default, XGBoost sets learning_rate=0.1\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "4609ac2a",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-10-10T14:20:32.899090Z",
     "iopub.status.busy": "2022-10-10T14:20:32.898663Z",
     "iopub.status.idle": "2022-10-10T14:20:34.699553Z",
     "shell.execute_reply": "2022-10-10T14:20:34.698638Z"
    },
    "papermill": {
     "duration": 1.812897,
     "end_time": "2022-10-10T14:20:34.702375",
     "exception": false,
     "start_time": "2022-10-10T14:20:32.889478",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "XGBRegressor(base_score=0.5, booster='gbtree', callbacks=None,\n",
       "             colsample_bylevel=1, colsample_bynode=1, colsample_bytree=1,\n",
       "             early_stopping_rounds=None, enable_categorical=False,\n",
       "             eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',\n",
       "             importance_type=None, interaction_constraints='',\n",
       "             learning_rate=0.05, max_bin=256, max_cat_to_onehot=4,\n",
       "             max_delta_step=0, max_depth=6, max_leaves=0, min_child_weight=1,\n",
       "             missing=nan, monotone_constraints='()', n_estimators=1000,\n",
       "             n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=0,\n",
       "             reg_alpha=0, reg_lambda=1, ...)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "my_model = XGBRegressor(n_estimators=1000, learning_rate=0.05)\n",
    "my_model.fit(X_train, y_train, \n",
    "             early_stopping_rounds=5, \n",
    "             eval_set=[(X_valid, y_valid)], \n",
    "             verbose=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "69cbaf0a",
   "metadata": {
    "papermill": {
     "duration": 0.007551,
     "end_time": "2022-10-10T14:20:34.717903",
     "exception": false,
     "start_time": "2022-10-10T14:20:34.710352",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "#### n_jobs\n",
    "\n",
    "On larger datasets where runtime is a consideration, you can use parallelism to build your models faster. It's common to set the parameter n_jobs equal to the number of cores on your machine. On smaller datasets, this won't help.\n",
    "\n",
    "The resulting model won't be any better, so micro-optimizing for fitting time is typically nothing but a distraction. But, it's useful in large datasets where you would otherwise spend a long time waiting during the fit command."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "8b2f30a0",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-10-10T14:20:34.735797Z",
     "iopub.status.busy": "2022-10-10T14:20:34.735403Z",
     "iopub.status.idle": "2022-10-10T14:20:36.555605Z",
     "shell.execute_reply": "2022-10-10T14:20:36.554420Z"
    },
    "papermill": {
     "duration": 1.832604,
     "end_time": "2022-10-10T14:20:36.558541",
     "exception": false,
     "start_time": "2022-10-10T14:20:34.725937",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "XGBRegressor(base_score=0.5, booster='gbtree', callbacks=None,\n",
       "             colsample_bylevel=1, colsample_bynode=1, colsample_bytree=1,\n",
       "             early_stopping_rounds=None, enable_categorical=False,\n",
       "             eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',\n",
       "             importance_type=None, interaction_constraints='',\n",
       "             learning_rate=0.05, max_bin=256, max_cat_to_onehot=4,\n",
       "             max_delta_step=0, max_depth=6, max_leaves=0, min_child_weight=1,\n",
       "             missing=nan, monotone_constraints='()', n_estimators=1000,\n",
       "             n_jobs=4, num_parallel_tree=1, predictor='auto', random_state=0,\n",
       "             reg_alpha=0, reg_lambda=1, ...)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "my_model = XGBRegressor(n_estimators=1000, learning_rate=0.05, n_jobs=4)\n",
    "my_model.fit(X_train, y_train, \n",
    "             early_stopping_rounds=5, \n",
    "             eval_set=[(X_valid, y_valid)], \n",
    "             verbose=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1fb33987",
   "metadata": {
    "papermill": {
     "duration": 0.008293,
     "end_time": "2022-10-10T14:20:36.575244",
     "exception": false,
     "start_time": "2022-10-10T14:20:36.566951",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Data Leakage\n",
    "\n",
    "Data leakage (or leakage) happens when your training data contains information about the target, but similar data will not be available when the model is used for prediction. This leads to high performance on the training set (and possibly even the validation data), but the model will perform poorly in production.\n",
    "\n",
    "There are two main types of leakage: **target leakage** and **train-test contamination.**\n",
    "\n",
    "#### Target leakage\n",
    "\n",
    "Target leakage occurs when your predictors include data that will not be available at the time you make predictions. It is important to think about target leakage in terms of the timing or chronological order that data becomes available, not merely whether a feature helps make good predictions.\n",
    "\n",
    "#### Train-Test contamination\n",
    "\n",
    "Recall that validation is meant to be a measure of how the model does on data that it hasn't considered before. You can corrupt this process in subtle ways if the validation data affects the preprocessing behavior. This is sometimes called train-test contamination.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "a22352ca",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-10-10T14:20:36.594265Z",
     "iopub.status.busy": "2022-10-10T14:20:36.593205Z",
     "iopub.status.idle": "2022-10-10T14:20:37.662183Z",
     "shell.execute_reply": "2022-10-10T14:20:37.660838Z"
    },
    "papermill": {
     "duration": 1.0817,
     "end_time": "2022-10-10T14:20:37.665201",
     "exception": false,
     "start_time": "2022-10-10T14:20:36.583501",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of rows in the dataset: 1319\n",
      "Cross-validation accuracy: 0.979534\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.model_selection import cross_val_score\n",
    "\n",
    "# Read the data\n",
    "data = pd.read_csv('../input/aer-credit-card-datacsv/AER_credit_card_data.csv', \n",
    "                   true_values = ['yes'], false_values = ['no'])\n",
    "\n",
    "# Select target\n",
    "y = data.card\n",
    "\n",
    "# Select predictors\n",
    "X = data.drop(['card'], axis=1)\n",
    "\n",
    "print(\"Number of rows in the dataset:\", X.shape[0])\n",
    "X.head()\n",
    "\n",
    "\n",
    "\n",
    "# Since there is no preprocessing, we don't need a pipeline (used anyway as best practice!)\n",
    "my_pipeline = make_pipeline(RandomForestClassifier(n_estimators=100))\n",
    "cv_scores = cross_val_score(my_pipeline, X, y, \n",
    "                            cv=5,\n",
    "                            scoring='accuracy')\n",
    "\n",
    "print(\"Cross-validation accuracy: %f\" % cv_scores.mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "f444a69a",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-10-10T14:20:37.683613Z",
     "iopub.status.busy": "2022-10-10T14:20:37.683228Z",
     "iopub.status.idle": "2022-10-10T14:20:38.970519Z",
     "shell.execute_reply": "2022-10-10T14:20:38.969260Z"
    },
    "papermill": {
     "duration": 1.299687,
     "end_time": "2022-10-10T14:20:38.973288",
     "exception": false,
     "start_time": "2022-10-10T14:20:37.673601",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cross-val accuracy: 0.830171\n"
     ]
    }
   ],
   "source": [
    "# Drop leaky predictors from dataset\n",
    "potential_leaks = ['expenditure', 'share', 'active', 'majorcards']\n",
    "X2 = X.drop(potential_leaks, axis=1)\n",
    "\n",
    "# Evaluate the model with leaky predictors removed\n",
    "cv_scores = cross_val_score(my_pipeline, X2, y, \n",
    "                            cv=5,\n",
    "                            scoring='accuracy')\n",
    "\n",
    "print(\"Cross-val accuracy: %f\" % cv_scores.mean())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6f22ef82",
   "metadata": {
    "papermill": {
     "duration": 0.008221,
     "end_time": "2022-10-10T14:20:38.989918",
     "exception": false,
     "start_time": "2022-10-10T14:20:38.981697",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "As you can see, eventhough accuracy dropped a bit, it will always be consistent above 80% by preventing data leakage."
   ]
  }
 ],
 "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": 37.232443,
   "end_time": "2022-10-10T14:20:39.821880",
   "environment_variables": {},
   "exception": null,
   "input_path": "__notebook__.ipynb",
   "output_path": "__notebook__.ipynb",
   "parameters": {},
   "start_time": "2022-10-10T14:20:02.589437",
   "version": "2.3.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}