COMP8260 – AI Systems Class 2.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# COMP8260 – AI Systems Class 2\n",
    "\n",
    "We are exploring classification on the adult dataset, which is designed to predict high\n",
    "income earners (>50k$) from US census information such as education, hours of\n",
    "work per week, etc. While approaching the exercises below it could be useful to keep the\n",
    "following pages open in your browser:  \n",
    "\n",
    "sklearn API https://scikit-learn.org/stable/modules/classes.html  \n",
    "\n",
    "pandas getting started tutorial\n",
    "https://pandas.pydata.org/docs/getting_started/intro_tutorials/index.html  \n",
    "\n",
    "pyplot tutorial https://matplotlib.org/stable/tutorials/introductory/pyplot.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/jupyterhub/pyvenv/lib/python3.10/site-packages/sklearn/datasets/_openml.py:1022: FutureWarning: The default value of `parser` will change from `'liac-arff'` to `'auto'` in 1.4. You can set `parser='auto'` to silence this warning. Therefore, an `ImportError` will be raised from 1.4 if the dataset is dense and pandas is not installed. Note that the pandas parser may return different data types. See the Notes Section in fetch_openml's API doc for details.\n",
      "  warn(\n"
     ]
    }
   ],
   "source": [
    "import sklearn as sk\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from sklearn import datasets\n",
    "\n",
    "# Load the adult dataset (version 2) from openml.org using the sklearn.datasets.fetch_openml function.\n",
    "# The dataset contains 48842 instances, 14 attributes and 2 classes (binary classification).\n",
    "# The data is stored in a pandas dataframe.\n",
    "adult = datasets.fetch_openml('adult', version=2, as_frame=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 48842 entries, 0 to 48841\n",
      "Data columns (total 14 columns):\n",
      " #   Column          Non-Null Count  Dtype   \n",
      "---  ------          --------------  -----   \n",
      " 0   age             48842 non-null  float64 \n",
      " 1   workclass       46043 non-null  category\n",
      " 2   fnlwgt          48842 non-null  float64 \n",
      " 3   education       48842 non-null  category\n",
      " 4   education-num   48842 non-null  float64 \n",
      " 5   marital-status  48842 non-null  category\n",
      " 6   occupation      46033 non-null  category\n",
      " 7   relationship    48842 non-null  category\n",
      " 8   race            48842 non-null  category\n",
      " 9   sex             48842 non-null  category\n",
      " 10  capital-gain    48842 non-null  float64 \n",
      " 11  capital-loss    48842 non-null  float64 \n",
      " 12  hours-per-week  48842 non-null  float64 \n",
      " 13  native-country  47985 non-null  category\n",
      "dtypes: category(8), float64(6)\n",
      "memory usage: 2.6 MB\n"
     ]
    }
   ],
   "source": [
    "# Print the type of each feature to see if they are \n",
    "# categorical or numeric (hint you can use pandas.Dataframe.info to get \n",
    "# information about columns in the DataFrame)\n",
    "adult.data.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(48842, 14)\n",
      "class\n",
      "<=50K    37155\n",
      ">50K     11687\n",
      "Name: count, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "# Print out the size of the dataset (number of instances) and the \n",
    "# distribution of the target across the two classes using \n",
    "# pandas.Series.value_counts\n",
    "print(adult.data.shape)\n",
    "print(adult.target.value_counts())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(32724, 14)\n",
      "(16118, 14)\n",
      "(32724,)\n",
      "(16118,)\n"
     ]
    }
   ],
   "source": [
    "# Split the data into a train datasets (X_train, y_train) and test dataset \n",
    "# (X_test, y_test), using the sklearn.model_selection. \n",
    "# train_test_split function. Verify the size of each dataset looking at the \n",
    "# shape attribute.\n",
    "from sklearn.model_selection import train_test_split\n",
    "X_train, X_test, y_train, y_test = train_test_split(adult.data, adult.target, test_size=0.33, random_state=42) \n",
    "print(X_train.shape)\n",
    "print(X_test.shape)\n",
    "print(y_train.shape)\n",
    "print(y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create an X_train_num training dataset by dropping the non-numerical features \n",
    "# from the input data, (Hint you can select the relevant columns using [] and a list of \n",
    "# column names, or use pandas.Dataframe.drop to drop the categorical \n",
    "# columns or use pandas.DataFrame.select_dtypes).\n",
    "X_train_num = X_train.select_dtypes(include=np.number)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9989610072118323\n",
      "0.7746618687182033\n",
      "57\n",
      "6651\n"
     ]
    }
   ],
   "source": [
    "# Train a DecisionTreeClassifier with default parameters to predict the \n",
    "# target class from the numerical attributes of input using its fit method. Compute the \n",
    "# accuracy of the classifier over the training data and the test data. Hint: you can use \n",
    "# the predict method of the classifier to obtain the predicted labels from the train \n",
    "# and test inputs and use sklearn.metrics.accuracy_score to compute the \n",
    "# accuracy. Does the decision tree seem to overfit? Why? Print out the depth and the \n",
    "# number of leaves for the tree.\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.metrics import accuracy_score\n",
    "dt = DecisionTreeClassifier()\n",
    "dt.fit(X_train_num, y_train)\n",
    "y_pred_train = dt.predict(X_train_num)\n",
    "y_pred_test = dt.predict(X_test.select_dtypes(include=np.number))\n",
    "print(accuracy_score(y_train, y_pred_train))\n",
    "print(accuracy_score(y_test, y_pred_test))\n",
    "print(dt.get_depth())\n",
    "print(dt.get_n_leaves())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot the training and testing accuracy vs the maximum tree depth.\n",
    "train_acc = []\n",
    "test_acc = []\n",
    "for i in range(1, 30):\n",
    "    dt = DecisionTreeClassifier(max_depth=i)\n",
    "    dt.fit(X_train_num, y_train)\n",
    "    y_pred_train = dt.predict(X_train_num)\n",
    "    y_pred_test = dt.predict(X_test.select_dtypes(include=np.number))\n",
    "    train_acc.append(accuracy_score(y_train, y_pred_train))\n",
    "    test_acc.append(accuracy_score(y_test, y_pred_test))\n",
    "plt.plot(range(1, 30), train_acc, label=\"Training Accuracy\")\n",
    "plt.plot(range(1, 30), test_acc, label=\"Testing Accuracy\")\n",
    "plt.xlabel(\"Max Depth\")\n",
    "plt.ylabel(\"Accuracy\")\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Change the plot above to show the average 3-fold training and validation score using \n",
    "# the cross_validate function. Hint: to return the training scores for each fold you need \n",
    "# to specify return_train_score=True\n",
    "from sklearn.model_selection import cross_validate\n",
    "train_acc = []\n",
    "test_acc = []\n",
    "for i in range(1, 30):\n",
    "    dt = DecisionTreeClassifier(max_depth=i)\n",
    "    scores = cross_validate(dt, X_train_num, y_train, cv=3, return_train_score=True)\n",
    "    train_acc.append(scores['train_score'].mean())\n",
    "    test_acc.append(scores['test_score'].mean())\n",
    "plt.plot(range(1, 30), train_acc, label=\"Training Accuracy\")\n",
    "plt.plot(range(1, 30), test_acc, label=\"Testing Accuracy\")\n",
    "plt.xlabel(\"Max Depth\")\n",
    "plt.ylabel(\"Accuracy\")\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'max_depth': 9, 'min_samples_leaf': 6, 'min_samples_split': 9}\n",
      "0.8337000366703337\n",
      "9\n",
      "164\n"
     ]
    }
   ],
   "source": [
    "# Use sklearn.model_selection.GridSearchCV to find the best combination of max_depth, \n",
    "# min_samples_split and min_samples_leaf to constrain the complexity of the tree. \n",
    "# Keep each parameter to max 3-4 choices or the computation would take a long \n",
    "# amount of time.\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "dt = DecisionTreeClassifier()\n",
    "params = {'max_depth': range(1, 10), 'min_samples_split': range(2, 10), 'min_samples_leaf': range(1, 10)}\n",
    "gs = GridSearchCV(dt, params, cv=3)\n",
    "gs.fit(X_train_num, y_train)\n",
    "print(gs.best_params_)\n",
    "print(gs.best_score_)\n",
    "print(gs.best_estimator_.get_depth())\n",
    "print(gs.best_estimator_.get_n_leaves())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.8644419997555312\n",
      "0.8226827149770443\n"
     ]
    }
   ],
   "source": [
    "# Obtain an X_train_cat by keeping only the categorical features from X_train. \n",
    "# Sklearn’s DecisionTreeClassifier implementation does not natively support learning \n",
    "# from categorical features. Use sklearn.preprocessing.OneHotEncoder to transform \n",
    "# X_train_cat into X_train_enc (using the fit and the transform method) and train a \n",
    "# DecisionTreeClassifier using it. Compare its test and training performance with the \n",
    "# decision tree computed using the numerical features.\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "X_train_cat = X_train.select_dtypes(exclude=np.number)\n",
    "enc = OneHotEncoder()\n",
    "enc.fit(X_train_cat)\n",
    "X_train_enc = enc.transform(X_train_cat)\n",
    "dt = DecisionTreeClassifier()\n",
    "dt.fit(X_train_enc, y_train)\n",
    "y_pred_train = dt.predict(X_train_enc)\n",
    "y_pred_test = dt.predict(enc.transform(X_test.select_dtypes(exclude=np.number)))\n",
    "print(accuracy_score(y_train, y_pred_train))\n",
    "print(accuracy_score(y_test, y_pred_test)) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "workclass         1888\n",
      "education            0\n",
      "marital-status       0\n",
      "occupation        1896\n",
      "relationship         0\n",
      "race                 0\n",
      "sex                  0\n",
      "native-country     563\n",
      "dtype: int64\n",
      "['workclass_Federal-gov' 'workclass_Local-gov' 'workclass_Never-worked'\n",
      " 'workclass_Private' 'workclass_Self-emp-inc' 'workclass_Self-emp-not-inc'\n",
      " 'workclass_State-gov' 'workclass_Without-pay' 'workclass_nan'\n",
      " 'education_10th' 'education_11th' 'education_12th' 'education_1st-4th'\n",
      " 'education_5th-6th' 'education_7th-8th' 'education_9th'\n",
      " 'education_Assoc-acdm' 'education_Assoc-voc' 'education_Bachelors'\n",
      " 'education_Doctorate' 'education_HS-grad' 'education_Masters'\n",
      " 'education_Preschool' 'education_Prof-school' 'education_Some-college'\n",
      " 'marital-status_Divorced' 'marital-status_Married-AF-spouse'\n",
      " 'marital-status_Married-civ-spouse'\n",
      " 'marital-status_Married-spouse-absent' 'marital-status_Never-married'\n",
      " 'marital-status_Separated' 'marital-status_Widowed'\n",
      " 'occupation_Adm-clerical' 'occupation_Armed-Forces'\n",
      " 'occupation_Craft-repair' 'occupation_Exec-managerial'\n",
      " 'occupation_Farming-fishing' 'occupation_Handlers-cleaners'\n",
      " 'occupation_Machine-op-inspct' 'occupation_Other-service'\n",
      " 'occupation_Priv-house-serv' 'occupation_Prof-specialty'\n",
      " 'occupation_Protective-serv' 'occupation_Sales' 'occupation_Tech-support'\n",
      " 'occupation_Transport-moving' 'occupation_nan' 'relationship_Husband'\n",
      " 'relationship_Not-in-family' 'relationship_Other-relative'\n",
      " 'relationship_Own-child' 'relationship_Unmarried' 'relationship_Wife'\n",
      " 'race_Amer-Indian-Eskimo' 'race_Asian-Pac-Islander' 'race_Black'\n",
      " 'race_Other' 'race_White' 'sex_Female' 'sex_Male'\n",
      " 'native-country_Cambodia' 'native-country_Canada' 'native-country_China'\n",
      " 'native-country_Columbia' 'native-country_Cuba'\n",
      " 'native-country_Dominican-Republic' 'native-country_Ecuador'\n",
      " 'native-country_El-Salvador' 'native-country_England'\n",
      " 'native-country_France' 'native-country_Germany' 'native-country_Greece'\n",
      " 'native-country_Guatemala' 'native-country_Haiti'\n",
      " 'native-country_Holand-Netherlands' 'native-country_Honduras'\n",
      " 'native-country_Hong' 'native-country_Hungary' 'native-country_India'\n",
      " 'native-country_Iran' 'native-country_Ireland' 'native-country_Italy'\n",
      " 'native-country_Jamaica' 'native-country_Japan' 'native-country_Laos'\n",
      " 'native-country_Mexico' 'native-country_Nicaragua'\n",
      " 'native-country_Outlying-US(Guam-USVI-etc)' 'native-country_Peru'\n",
      " 'native-country_Philippines' 'native-country_Poland'\n",
      " 'native-country_Portugal' 'native-country_Puerto-Rico'\n",
      " 'native-country_Scotland' 'native-country_South' 'native-country_Taiwan'\n",
      " 'native-country_Thailand' 'native-country_Trinadad&Tobago'\n",
      " 'native-country_United-States' 'native-country_Vietnam'\n",
      " 'native-country_Yugoslavia' 'native-country_nan']\n"
     ]
    }
   ],
   "source": [
    "# Did X_train_cat contain any missing values? Can you figure out what happened to \n",
    "# the missing values during the encoding by looking at the output of the \n",
    "# get_feature_names_out() method of the encoder?\n",
    "print(X_train_cat.isna().sum())\n",
    "print(enc.get_feature_names_out())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.8619361936193619\n",
      "0.8211316540513711\n",
      "['x0_Federal-gov' 'x0_Local-gov' 'x0_Never-worked' 'x0_Private'\n",
      " 'x0_Self-emp-inc' 'x0_Self-emp-not-inc' 'x0_State-gov' 'x0_Without-pay'\n",
      " 'x1_10th' 'x1_11th' 'x1_12th' 'x1_1st-4th' 'x1_5th-6th' 'x1_7th-8th'\n",
      " 'x1_9th' 'x1_Assoc-acdm' 'x1_Assoc-voc' 'x1_Bachelors' 'x1_Doctorate'\n",
      " 'x1_HS-grad' 'x1_Masters' 'x1_Preschool' 'x1_Prof-school'\n",
      " 'x1_Some-college' 'x2_Divorced' 'x2_Married-AF-spouse'\n",
      " 'x2_Married-civ-spouse' 'x2_Married-spouse-absent' 'x2_Never-married'\n",
      " 'x2_Separated' 'x2_Widowed' 'x3_Adm-clerical' 'x3_Armed-Forces'\n",
      " 'x3_Craft-repair' 'x3_Exec-managerial' 'x3_Farming-fishing'\n",
      " 'x3_Handlers-cleaners' 'x3_Machine-op-inspct' 'x3_Other-service'\n",
      " 'x3_Priv-house-serv' 'x3_Prof-specialty' 'x3_Protective-serv' 'x3_Sales'\n",
      " 'x3_Tech-support' 'x3_Transport-moving' 'x4_Husband' 'x4_Not-in-family'\n",
      " 'x4_Other-relative' 'x4_Own-child' 'x4_Unmarried' 'x4_Wife'\n",
      " 'x5_Amer-Indian-Eskimo' 'x5_Asian-Pac-Islander' 'x5_Black' 'x5_Other'\n",
      " 'x5_White' 'x6_Female' 'x6_Male' 'x7_Cambodia' 'x7_Canada' 'x7_China'\n",
      " 'x7_Columbia' 'x7_Cuba' 'x7_Dominican-Republic' 'x7_Ecuador'\n",
      " 'x7_El-Salvador' 'x7_England' 'x7_France' 'x7_Germany' 'x7_Greece'\n",
      " 'x7_Guatemala' 'x7_Haiti' 'x7_Holand-Netherlands' 'x7_Honduras' 'x7_Hong'\n",
      " 'x7_Hungary' 'x7_India' 'x7_Iran' 'x7_Ireland' 'x7_Italy' 'x7_Jamaica'\n",
      " 'x7_Japan' 'x7_Laos' 'x7_Mexico' 'x7_Nicaragua'\n",
      " 'x7_Outlying-US(Guam-USVI-etc)' 'x7_Peru' 'x7_Philippines' 'x7_Poland'\n",
      " 'x7_Portugal' 'x7_Puerto-Rico' 'x7_Scotland' 'x7_South' 'x7_Taiwan'\n",
      " 'x7_Thailand' 'x7_Trinadad&Tobago' 'x7_United-States' 'x7_Vietnam'\n",
      " 'x7_Yugoslavia']\n"
     ]
    }
   ],
   "source": [
    "# Use SimpleImputer to fill in the missing values in X_train_cat using the most frequent value of the \n",
    "# missing attributes before encoding. Print the set of encoded features \n",
    "# and the classification performance\n",
    "from sklearn.impute import SimpleImputer\n",
    "imp = SimpleImputer(strategy='most_frequent')\n",
    "# Fit the SimpleImputer on the entire dataset\n",
    "imp.fit(pd.concat([X_train_cat, X_test.select_dtypes(exclude=np.number)]))\n",
    "\n",
    "# Transform the training and test data\n",
    "X_train_cat_imp = pd.DataFrame(imp.transform(X_train_cat))\n",
    "X_test_cat_imp = pd.DataFrame(imp.transform(X_test.select_dtypes(exclude=np.number)))\n",
    "\n",
    "# Fit the OneHotEncoder on the imputed dataset\n",
    "enc.fit(pd.concat([X_train_cat_imp, X_test_cat_imp]))\n",
    "\n",
    "# Transform the training and test data\n",
    "X_train_enc = enc.transform(X_train_cat_imp)\n",
    "X_test_enc = enc.transform(X_test_cat_imp)\n",
    "\n",
    "# Continue with the rest of your code...\n",
    "dt = DecisionTreeClassifier()\n",
    "dt.fit(X_train_enc, y_train)\n",
    "y_pred_train = dt.predict(X_train_enc)\n",
    "y_pred_test = dt.predict(X_test_enc)\n",
    "print(accuracy_score(y_train, y_pred_train))\n",
    "print(accuracy_score(y_test, y_pred_test))\n",
    "print(enc.get_feature_names_out())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.8619361936193619\n",
      "0.8213177813624519\n"
     ]
    }
   ],
   "source": [
    "# Automate the use of SimpleImputer and OneHotEncoder by creating a Pipeline. \n",
    "# Create a Pipeline with a SimpleImputer, OneHotEncoder and a \n",
    "# DecisionTreeClassifier. Train and verify the accuracy of the pipeline.\n",
    "from sklearn.pipeline import Pipeline\n",
    "pipe = Pipeline([('imputer', SimpleImputer(strategy='most_frequent')), ('encoder', OneHotEncoder()), ('dt', DecisionTreeClassifier())])\n",
    "pipe.fit(X_train_cat, y_train)\n",
    "y_pred_train = pipe.predict(X_train_cat)\n",
    "y_pred_test = pipe.predict(X_test.select_dtypes(exclude=np.number))\n",
    "print(accuracy_score(y_train, y_pred_train))\n",
    "print(accuracy_score(y_test, y_pred_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9999083241657499\n",
      "0.8208834843032634\n"
     ]
    }
   ],
   "source": [
    "#  It would be nice see if a Decision Tree classifier would be able to reach a higher \n",
    "# accuracy by considering both categorical and numerical features. To do so we could \n",
    "# start from the full dataset X_train and apply a transformer to only a subset of the \n",
    "# attributes (DataFrame columns). Use a ColumnTransformer to apply the categorical \n",
    "# pre-processing pipeline (composed by an Imputer and one OneHotEncoder) to the \n",
    "# categorical attributes of the data, leaving the numerical attribute unaltered (hint: \n",
    "# have a look at the remainder parameter of ColumnTransformer).\n",
    "from sklearn.compose import ColumnTransformer\n",
    "pipe = Pipeline([('imputer', SimpleImputer(strategy='most_frequent')), ('encoder', OneHotEncoder())])\n",
    "ct = ColumnTransformer([('cat', pipe, X_train_cat.columns)], remainder='passthrough')\n",
    "ct.fit(X_train)\n",
    "X_train_enc = ct.transform(X_train)\n",
    "X_test_enc = ct.transform(X_test)\n",
    "dt = DecisionTreeClassifier()\n",
    "dt.fit(X_train_enc, y_train)\n",
    "y_pred_train = dt.predict(X_train_enc)\n",
    "y_pred_test = dt.predict(X_test_enc)\n",
    "print(accuracy_score(y_train, y_pred_train))\n",
    "print(accuracy_score(y_test, y_pred_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9999083241657499\n",
      "0.8206973569921826\n"
     ]
    }
   ],
   "source": [
    "# Define a Pipeline composed of the ColumnTransformer above and a \n",
    "# DecisionTreeClassifier and train and verify the accuracy of the Pipeline. Does the use \n",
    "# of both types of features improve the accuracy?\n",
    "pipe = Pipeline([('ct', ct), ('dt', DecisionTreeClassifier())])\n",
    "pipe.fit(X_train, y_train)\n",
    "y_pred_train = pipe.predict(X_train)\n",
    "y_pred_test = pipe.predict(X_test)\n",
    "print(accuracy_score(y_train, y_pred_train))\n",
    "print(accuracy_score(y_test, y_pred_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9998777655543332\n",
      "0.8594738801340117\n",
      "65\n",
      "5876\n"
     ]
    }
   ],
   "source": [
    "# Train a RandomForestClassifier on the full dataset, modifying the pipeline of the \n",
    "# previous point. Compare the training and testing accuracy with the previous cases: \n",
    "# does the classifier overfit the data? How does the accuracy compare with the \n",
    "# previous DecisionTrees? Print out the size of each tree in the ensemble\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "pipe = Pipeline([('ct', ct), ('rf', RandomForestClassifier())])\n",
    "pipe.fit(X_train, y_train)\n",
    "y_pred_train = pipe.predict(X_train)\n",
    "y_pred_test = pipe.predict(X_test)\n",
    "print(accuracy_score(y_train, y_pred_train))\n",
    "print(accuracy_score(y_test, y_pred_test))\n",
    "print(pipe['rf'].estimators_[0].get_depth())\n",
    "print(pipe['rf'].estimators_[0].get_n_leaves())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.8579330155237745\n",
      "0.8659883360218389\n",
      "1\n",
      "2\n",
      "0.7590453489793424\n",
      "0.7641146544236258\n",
      "2\n",
      "4\n"
     ]
    }
   ],
   "source": [
    "# Train an AdaBoostClassifier on the dataset. Compare train and test accuracy with \n",
    "# random forest print out the size of each tree in the ensemble. Compare the \n",
    "# performance of the classifier with a RandomForestClassifier that uses the same \n",
    "# number of trees and max_depth as the AdaBoostClassifier.\n",
    "from sklearn.ensemble import AdaBoostClassifier\n",
    "pipe = Pipeline([('ct', ct), ('ada', AdaBoostClassifier())])\n",
    "pipe.fit(X_train, y_train)\n",
    "y_pred_train = pipe.predict(X_train)\n",
    "y_pred_test = pipe.predict(X_test)\n",
    "print(accuracy_score(y_train, y_pred_train))\n",
    "print(accuracy_score(y_test, y_pred_test))\n",
    "print(pipe['ada'].estimators_[0].get_depth())\n",
    "print(pipe['ada'].estimators_[0].get_n_leaves())\n",
    "pipe = Pipeline([('ct', ct), ('rf', RandomForestClassifier(n_estimators=1, max_depth=2))])\n",
    "pipe.fit(X_train, y_train)\n",
    "y_pred_train = pipe.predict(X_train)\n",
    "y_pred_test = pipe.predict(X_test)\n",
    "print(accuracy_score(y_train, y_pred_train))\n",
    "print(accuracy_score(y_test, y_pred_test))\n",
    "print(pipe['rf'].estimators_[0].get_depth())\n",
    "print(pipe['rf'].estimators_[0].get_n_leaves())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.8674673022857842\n",
      "0.8723787070356124\n",
      "3\n",
      "8\n"
     ]
    }
   ],
   "source": [
    "# Train a GradientBoostingClassifier on the dataset and compare the performance \n",
    "# with the above.\n",
    "from sklearn.ensemble import GradientBoostingClassifier\n",
    "pipe = Pipeline([('ct', ct), ('gb', GradientBoostingClassifier())])\n",
    "pipe.fit(X_train, y_train)\n",
    "y_pred_train = pipe.predict(X_train)\n",
    "y_pred_test = pipe.predict(X_test)\n",
    "print(accuracy_score(y_train, y_pred_train))\n",
    "print(accuracy_score(y_test, y_pred_test))\n",
    "print(pipe['gb'].estimators_[0][0].get_depth())\n",
    "print(pipe['gb'].estimators_[0][0].get_n_leaves())  "
   ]
  }
 ],
 "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": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}