{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Decison Trees"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"First we'll load some fake data on past hires I made up. Note how we use pandas to convert a csv file into a DataFrame:"
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
"import numpy as np\n",
"import pandas as pd\n",
"from sklearn import tree\n",
"\n",
"input_file = \"PastHires.csv\"\n",
"df = pd.read_csv(input_file, header = 0)"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"<div>\n",
"<style scoped>\n",
" .dataframe tbody tr th:only-of-type {\n",
" vertical-align: middle;\n",
" }\n",
"\n",
" .dataframe tbody tr th {\n",
" vertical-align: top;\n",
" }\n",
"\n",
" .dataframe thead th {\n",
" text-align: right;\n",
" }\n",
"</style>\n",
"<table border=\"1\" class=\"dataframe\">\n",
" <thead>\n",
" <tr style=\"text-align: right;\">\n",
" <th></th>\n",
" <th>Years Experience</th>\n",
" <th>Employed?</th>\n",
" <th>Previous employers</th>\n",
" <th>Level of Education</th>\n",
" <th>Top-tier school</th>\n",
" <th>Interned</th>\n",
" <th>Hired</th>\n",
" </tr>\n",
" </thead>\n",
" <tbody>\n",
" <tr>\n",
" <th>0</th>\n",
" <td>10</td>\n",
" <td>Y</td>\n",
" <td>4</td>\n",
" <td>BS</td>\n",
" <td>N</td>\n",
" <td>N</td>\n",
" <td>Y</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1</th>\n",
" <td>0</td>\n",
" <td>N</td>\n",
" <td>0</td>\n",
" <td>BS</td>\n",
" <td>Y</td>\n",
" <td>Y</td>\n",
" <td>Y</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2</th>\n",
" <td>7</td>\n",
" <td>N</td>\n",
" <td>6</td>\n",
" <td>BS</td>\n",
" <td>N</td>\n",
" <td>N</td>\n",
" <td>N</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3</th>\n",
" <td>2</td>\n",
" <td>Y</td>\n",
" <td>1</td>\n",
" <td>MS</td>\n",
" <td>Y</td>\n",
" <td>N</td>\n",
" <td>Y</td>\n",
" </tr>\n",
" <tr>\n",
" <th>4</th>\n",
" <td>20</td>\n",
" <td>N</td>\n",
" <td>2</td>\n",
" <td>PhD</td>\n",
" <td>Y</td>\n",
" <td>N</td>\n",
" <td>N</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
"</div>"
],
"text/plain": [
" Years Experience Employed? Previous employers Level of Education \\\n",
"0 10 Y 4 BS \n",
"1 0 N 0 BS \n",
"2 7 N 6 BS \n",
"3 2 Y 1 MS \n",
"4 20 N 2 PhD \n",
"\n",
" Top-tier school Interned Hired \n",
"0 N N Y \n",
"1 Y Y Y \n",
"2 N N N \n",
"3 Y N Y \n",
"4 Y N N "
]
},
"execution_count": 2,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"df.head()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"scikit-learn needs everything to be numerical for decision trees to work. So, we'll map Y,N to 1,0 and levels of education to some scale of 0-2. In the real world, you'd need to think about how to deal with unexpected or missing data! By using map(), we know we'll get NaN for unexpected values."
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"<div>\n",
"<style scoped>\n",
" .dataframe tbody tr th:only-of-type {\n",
" vertical-align: middle;\n",
" }\n",
"\n",
" .dataframe tbody tr th {\n",
" vertical-align: top;\n",
" }\n",
"\n",
" .dataframe thead th {\n",
" text-align: right;\n",
" }\n",
"</style>\n",
"<table border=\"1\" class=\"dataframe\">\n",
" <thead>\n",
" <tr style=\"text-align: right;\">\n",
" <th></th>\n",
" <th>Years Experience</th>\n",
" <th>Employed?</th>\n",
" <th>Previous employers</th>\n",
" <th>Level of Education</th>\n",
" <th>Top-tier school</th>\n",
" <th>Interned</th>\n",
" <th>Hired</th>\n",
" </tr>\n",
" </thead>\n",
" <tbody>\n",
" <tr>\n",
" <th>0</th>\n",
" <td>10</td>\n",
" <td>1</td>\n",
" <td>4</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1</th>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2</th>\n",
" <td>7</td>\n",
" <td>0</td>\n",
" <td>6</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3</th>\n",
" <td>2</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" </tr>\n",
" <tr>\n",
" <th>4</th>\n",
" <td>20</td>\n",
" <td>0</td>\n",
" <td>2</td>\n",
" <td>2</td>\n",
" <td>1</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
"</div>"
],
"text/plain": [
" Years Experience Employed? Previous employers Level of Education \\\n",
"0 10 1 4 0 \n",
"1 0 0 0 0 \n",
"2 7 0 6 0 \n",
"3 2 1 1 1 \n",
"4 20 0 2 2 \n",
"\n",
" Top-tier school Interned Hired \n",
"0 0 0 1 \n",
"1 1 1 1 \n",
"2 0 0 0 \n",
"3 1 0 1 \n",
"4 1 0 0 "
]
},
"execution_count": 3,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"d = {'Y': 1, 'N': 0}\n",
"df['Hired'] = df['Hired'].map(d)\n",
"df['Employed?'] = df['Employed?'].map(d)\n",
"df['Top-tier school'] = df['Top-tier school'].map(d)\n",
"df['Interned'] = df['Interned'].map(d)\n",
"d = {'BS': 0, 'MS': 1, 'PhD': 2}\n",
"df['Level of Education'] = df['Level of Education'].map(d)\n",
"df.head()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Next we need to separate the features from the target column that we're trying to bulid a decision tree for."
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"['Years Experience',\n",
" 'Employed?',\n",
" 'Previous employers',\n",
" 'Level of Education',\n",
" 'Top-tier school',\n",
" 'Interned']"
]
},
"execution_count": 4,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"features = list(df.columns[:6])\n",
"features"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now actually construct the decision tree:"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {},
"outputs": [],
"source": [
"y = df[\"Hired\"]\n",
"X = df[features]\n",
"clf = tree.DecisionTreeClassifier()\n",
"clf = clf.fit(X,y)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"... and display it. Note you need to have pydotplus installed for this to work. (!pip install pydotplus)\n",
"\n",
"To read this decision tree, each condition branches left for \"true\" and right for \"false\". When you end up at a value, the value array represents how many samples exist in each target value. So value = [0. 5.] mean there are 0 \"no hires\" and 5 \"hires\" by the tim we get to that point. value = [3. 0.] means 3 no-hires and 0 hires."
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<IPython.core.display.Image object>"
]
},
"execution_count": 6,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"from IPython.display import Image \n",
"from io import StringIO \n",
"import pydotplus\n",
"\n",
"dot_data = StringIO() \n",
"tree.export_graphviz(clf, out_file=dot_data, \n",
" feature_names=features) \n",
"graph = pydotplus.graph_from_dot_data(dot_data.getvalue()) \n",
"Image(graph.create_png()) "
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Ensemble learning: using a random forest"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We'll use a random forest of 10 decision trees to predict employment of specific candidate profiles:"
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"[1]\n",
"[1]\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"C:\\Users\\fkane\\anaconda3\\lib\\site-packages\\sklearn\\base.py:450: UserWarning: X does not have valid feature names, but RandomForestClassifier was fitted with feature names\n",
" warnings.warn(\n",
"C:\\Users\\fkane\\anaconda3\\lib\\site-packages\\sklearn\\base.py:450: UserWarning: X does not have valid feature names, but RandomForestClassifier was fitted with feature names\n",
" warnings.warn(\n"
]
}
],
"source": [
"from sklearn.ensemble import RandomForestClassifier\n",
"\n",
"clf = RandomForestClassifier(n_estimators=10)\n",
"clf = clf.fit(X, y)\n",
"\n",
"#Predict employment of an employed 10-year veteran\n",
"print (clf.predict([[10, 1, 4, 0, 0, 0]]))\n",
"#...and an unemployed 10-year veteran\n",
"print (clf.predict([[10, 0, 4, 0, 0, 0]]))"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Activity"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Modify the test data to create an alternate universe where everyone I hire everyone I normally wouldn't have, and vice versa. Compare the resulting decision tree to the one from the original data."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"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": "3.9.13"
},
"widgets": {
"state": {},
"version": "1.1.2"
}
},
"nbformat": 4,
"nbformat_minor": 1
}