{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Decision Trees Exercises"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Introduction\n",
"\n",
"We will be using the wine quality data set for these exercises. This data set contains various chemical properties of wine, such as acidity, sugar, pH, and alcohol. It also contains a quality metric (3-9, with highest being better) and a color (red or white). The name of the file is `Wine_Quality_Data.csv`."
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {
"ExecuteTime": {
"end_time": "2017-04-10T00:04:57.164238Z",
"start_time": "2017-04-09T20:04:57.158472-04:00"
}
},
"outputs": [],
"source": [
"from __future__ import print_function\n",
"import os\n",
"data_path = ['..', '..', 'data']"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Question 1\n",
"\n",
"* Import the data and examine the features.\n",
"* We will be using all of them to predict `color` (white or red), but the colors feature will need to be integer encoded."
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [],
"source": [
"import pandas as pd\n",
"path = 'data/Wine_Quality_data.csv'\n",
"data = pd.read_csv(path, sep=',')"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"fixed_acidity float64\n",
"volatile_acidity float64\n",
"citric_acid float64\n",
"residual_sugar float64\n",
"chlorides float64\n",
"free_sulfur_dioxide float64\n",
"total_sulfur_dioxide float64\n",
"density float64\n",
"pH float64\n",
"sulphates float64\n",
"alcohol float64\n",
"quality int64\n",
"color object\n",
"dtype: object"
]
},
"execution_count": 3,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"data.dtypes"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [],
"source": [
"data['color'] = data.color.replace('white', 0)\n",
"data['color'] = data.color.replace('red', 1)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"scrolled": true
},
"outputs": [],
"source": [
"print(data.color)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Question 2\n",
"\n",
"* Use `StratifiedShuffleSplit` to split data into train and test sets that are stratified by wine quality. If possible, preserve the indices of the split for question 5 below.\n",
"* Check the percent composition of each quality level for both the train and test data sets."
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"['fixed_acidity', 'volatile_acidity', 'citric_acid', 'residual_sugar', 'chlorides', 'free_sulfur_dioxide', 'total_sulfur_dioxide', 'density', 'pH', 'sulphates', 'alcohol', 'quality']\n"
]
}
],
"source": [
"features_cols = [x for x in data.columns if x not in 'color']\n",
"print(features_cols)"
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"5497\n",
"1000\n"
]
}
],
"source": [
"from sklearn.model_selection import StratifiedShuffleSplit\n",
"\n",
"splitter = StratifiedShuffleSplit(n_splits = 1, test_size=1000, random_state=40)\n",
"list_split = splitter.split(data[features_cols], data['color'])\n",
"train_idx, test_idx = next(list_split)\n",
"print(train_idx.size)\n",
"print(test_idx.size)"
]
},
{
"cell_type": "code",
"execution_count": 8,
"metadata": {},
"outputs": [],
"source": [
"X_train = data.loc[train_idx, features_cols]\n",
"y_train = data.loc[train_idx, 'color']\n",
"\n",
"X_test = data.loc[test_idx, features_cols]\n",
"y_test = data.loc[test_idx, 'color']"
]
},
{
"cell_type": "code",
"execution_count": 9,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"0 0.754\n",
"1 0.246\n",
"Name: color, dtype: float64"
]
},
"execution_count": 9,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"y_train.value_counts(normalize=True).sort_index()\n",
"y_test.value_counts(normalize=True).sort_index()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Question 3\n",
"\n",
"* Fit a decision tree classifier with no set limits on maximum depth, features, or leaves.\n",
"* Determine how many nodes are present and what the depth of this (very large) tree is.\n",
"* Using this tree, measure the prediction error in the train and test data sets. What do you think is going on here based on the differences in prediction error?"
]
},
{
"cell_type": "code",
"execution_count": 10,
"metadata": {},
"outputs": [],
"source": [
"from sklearn.tree import DecisionTreeClassifier\n",
"criterion = 'gini'\n",
"dt = DecisionTreeClassifier(criterion=criterion, max_depth=6)\n",
"dt = dt.fit(X_train, y_train)"
]
},
{
"cell_type": "code",
"execution_count": 11,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"(6, 57)"
]
},
"execution_count": 11,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"dt.tree_.max_depth, dt.tree_.node_count"
]
},
{
"cell_type": "code",
"execution_count": 12,
"metadata": {},
"outputs": [],
"source": [
"from sklearn.metrics import accuracy_score, recall_score, precision_score, f1_score\n",
"\n",
"def measure_score(y_true, y_pred, label):\n",
" df = pd.Series({'accuracy': accuracy_score(y_true, y_pred),\n",
" 'recall': accuracy_score(y_true, y_pred),\n",
" 'precision': accuracy_score(y_true, y_pred),\n",
" 'f1_score': accuracy_score(y_true, y_pred)}, name=label)\n",
" return df"
]
},
{
"cell_type": "code",
"execution_count": 13,
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"\n",
"\n",
"
\n",
" \n",
" \n",
" | \n",
" train | \n",
" test | \n",
"
\n",
" \n",
" \n",
" \n",
" | accuracy | \n",
" 0.992905 | \n",
" 0.986 | \n",
"
\n",
" \n",
" | f1_score | \n",
" 0.992905 | \n",
" 0.986 | \n",
"
\n",
" \n",
" | precision | \n",
" 0.992905 | \n",
" 0.986 | \n",
"
\n",
" \n",
" | recall | \n",
" 0.992905 | \n",
" 0.986 | \n",
"
\n",
" \n",
"
\n",
"