RandomForest.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Implementing Random Forest in Python"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### By: Justin Nunez\n",
    "#### Statistical Machine Learning Spring 2018"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "______________________________________________________________________________________\n",
    "### Data that will be used is 2016 weather data from NOAA Climate Data Online tool\n",
    "#### We will work with temperature data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "_____________________________"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Python"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# importing our libraries\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import seaborn as sb\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "features = pd.read_csv(\"/Users/justinnunez/Downloads/temps.csv\")\n",
    "features.head(15);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Following are explanations of the columns:\n",
    "#### year: 2016 for all data points\n",
    "#### month: number for month of the year\n",
    "#### day: number for day of the year\n",
    "#### week: day of the week as a character string\n",
    "#### temp_2: max temperature 2 days prior\n",
    "#### temp_1: max temperature 1 day prior\n",
    "#### average: historical average max temperature\n",
    "#### actual: max temperature measurement\n",
    "#### friend: your friend’s prediction, a random number between 20 below the average and 20 above the average"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# we will look at the shape of our data\n",
    "\n",
    "features.shape;\n",
    "\n",
    "# does not correlate with 366 days in 2016, which shows missing data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "features.columns;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# check for anomalies\n",
    "features.describe();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 348 entries, 0 to 347\n",
      "Data columns (total 12 columns):\n",
      "year              348 non-null int64\n",
      "month             348 non-null int64\n",
      "day               348 non-null int64\n",
      "week              348 non-null object\n",
      "temp_2            348 non-null int64\n",
      "temp_1            348 non-null int64\n",
      "average           348 non-null float64\n",
      "actual            348 non-null int64\n",
      "forecast_noaa     348 non-null int64\n",
      "forecast_acc      348 non-null int64\n",
      "forecast_under    348 non-null int64\n",
      "friend            348 non-null int64\n",
      "dtypes: float64(1), int64(10), object(1)\n",
      "memory usage: 32.7+ KB\n"
     ]
    }
   ],
   "source": [
    "# check for variable types: quantitative and qualitative\n",
    "features.info();"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## For categorical variables we will perform one hot encoding\n",
    "## which creates dummy variables for us without hard-coding it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>average</th>\n",
       "      <th>actual</th>\n",
       "      <th>forecast_noaa</th>\n",
       "      <th>forecast_acc</th>\n",
       "      <th>forecast_under</th>\n",
       "      <th>friend</th>\n",
       "      <th>week_Fri</th>\n",
       "      <th>week_Mon</th>\n",
       "      <th>week_Sat</th>\n",
       "      <th>week_Sun</th>\n",
       "      <th>week_Thurs</th>\n",
       "      <th>week_Tues</th>\n",
       "      <th>week_Wed</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>45.6</td>\n",
       "      <td>45</td>\n",
       "      <td>43</td>\n",
       "      <td>50</td>\n",
       "      <td>44</td>\n",
       "      <td>29</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>45.7</td>\n",
       "      <td>44</td>\n",
       "      <td>41</td>\n",
       "      <td>50</td>\n",
       "      <td>44</td>\n",
       "      <td>61</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>45.8</td>\n",
       "      <td>41</td>\n",
       "      <td>43</td>\n",
       "      <td>46</td>\n",
       "      <td>47</td>\n",
       "      <td>56</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>45.9</td>\n",
       "      <td>40</td>\n",
       "      <td>44</td>\n",
       "      <td>48</td>\n",
       "      <td>46</td>\n",
       "      <td>53</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>46.0</td>\n",
       "      <td>44</td>\n",
       "      <td>46</td>\n",
       "      <td>46</td>\n",
       "      <td>46</td>\n",
       "      <td>41</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   average  actual  forecast_noaa  forecast_acc  forecast_under  friend  \\\n",
       "0     45.6      45             43            50              44      29   \n",
       "1     45.7      44             41            50              44      61   \n",
       "2     45.8      41             43            46              47      56   \n",
       "3     45.9      40             44            48              46      53   \n",
       "4     46.0      44             46            46              46      41   \n",
       "\n",
       "   week_Fri  week_Mon  week_Sat  week_Sun  week_Thurs  week_Tues  week_Wed  \n",
       "0         1         0         0         0           0          0         0  \n",
       "1         0         0         1         0           0          0         0  \n",
       "2         0         0         0         1           0          0         0  \n",
       "3         0         1         0         0           0          0         0  \n",
       "4         0         0         0         0           0          1         0  "
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# One-hot encode the data using pandas get_dummies\n",
    "features = pd.get_dummies(features)\n",
    "features.iloc[:,5:].head(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### We want to predict \"Actual\" temperature"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "labels = np.array(features['actual'])\n",
    "# Separate it from our data now"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# We want to separate our predictor variable and turn it into a numpy array\n",
    "features = features.drop('actual', axis = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# save variable names into a list\n",
    "feature_list = list(features.columns)\n",
    "\n",
    "# Convert to numpy array\n",
    "features = np.array(features)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Testing and Training data!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* When it comes time to evaluate the model, we make predictions on the testing data where it only has access to the features. Since we have the actual answers for the test set, we can compare these predictions to the true value to judge how accurate the model is."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Using Skicit-learn to split data into training and testing sets\n",
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Split the data into training and testing sets\n",
    "train_features, test_features, train_labels, test_labels = train_test_split(features, labels, test_size = 0.25, random_state = 42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Check the shape\n",
    "train_features.shape;\n",
    "train_labels.shape;\n",
    "test_features.shape;\n",
    "test_labels.shape;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### What is our baseline model ? Our model if we were to just take educated guesses"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# The baseline predictions are the historical averages\n",
    "baseline_preds = test_features[:, feature_list.index('average')]\n",
    "# Baseline errors, and display average baseline error\n",
    "baseline_errors = abs(baseline_preds - test_labels)\n",
    "round(np.mean(baseline_errors), 2);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model Fit"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Algorithm\n",
    "from sklearn.ensemble import RandomForestRegressor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestRegressor(bootstrap=True, criterion='mse', max_depth=None,\n",
       "           max_features='auto', max_leaf_nodes=None,\n",
       "           min_impurity_split=1e-07, min_samples_leaf=1,\n",
       "           min_samples_split=2, min_weight_fraction_leaf=0.0,\n",
       "           n_estimators=1000, n_jobs=1, oob_score=False, random_state=42,\n",
       "           verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#1000 decision trees\n",
    "rf = RandomForestRegressor(n_estimators = 1000, random_state = 42)\n",
    "\n",
    "# Train the model on training data\n",
    "rf.fit(train_features, train_labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "predictions = rf.predict(test_features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Mean Absolute Error:', 3.87, 'degrees.')\n"
     ]
    }
   ],
   "source": [
    "# Calculate the absolute errors\n",
    "errors = abs(predictions - test_labels)\n",
    "\n",
    "# Print out the mean absolute error (mae)\n",
    "print('Mean Absolute Error:', round(np.mean(errors), 2), 'degrees.')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Accuracy:', 93.94, '%.')\n"
     ]
    }
   ],
   "source": [
    "# Calculate mean absolute percentage error (MAPE)\n",
    "mape = 100 * (errors / test_labels)\n",
    "# Calculate and display accuracy\n",
    "accuracy = 100 - np.mean(mape)\n",
    "print('Accuracy:', round(accuracy, 2), '%.')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Which Variables are important!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Get numerical feature importances\n",
    "importances = list(rf.feature_importances_)\n",
    "\n",
    "# List of tuples with variable and importance\n",
    "feature_importances = [(feature, round(importance, 2)) for feature, importance in zip(feature_list, importances)]\n",
    "\n",
    "# Sort the feature importances by most important first\n",
    "feature_importances = sorted(feature_importances, key = lambda x: x[1], reverse = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": false
   },
   "source": [
    "* Variable: temp_1               Importance: 0.7\n",
    "* Variable: average              Importance: 0.19\n",
    "* Variable: day                  Importance: 0.03\n",
    "* Variable: temp_2               Importance: 0.02\n",
    "* Variable: friend               Importance: 0.02\n",
    "* Variable: month                Importance: 0.01\n",
    "* Variable: year                 Importance: 0.0\n",
    "* Variable: week_Fri             Importance: 0.0\n",
    "* Variable: week_Mon             Importance: 0.0\n",
    "* Variable: week_Sat             Importance: 0.0\n",
    "* Variable: week_Sun             Importance: 0.0\n",
    "* Variable: week_Thurs           Importance: 0.0\n",
    "* Variable: week_Tues            Importance: 0.0\n",
    "* Variable: week_Wed             Importance: 0.0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### At the top of the list is temp_1, the max temperature of the day before. This tells us the best predictor of the max temperature for a day is the max temperature of the day before, a rather intuitive finding. The second most important factor is the historical average max temperature, also not that surprising."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAa4AAAFqCAYAAABCjL44AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nFFwiIpJQ/gGGK2AdaAuR6gAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11b862910>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Set the style\n",
    "plt.style.use('fivethirtyeight')\n",
    "# list of x locations for plotting\n",
    "x_values = list(range(len(importances)))\n",
    "# Make a bar chart\n",
    "plt.bar(x_values, importances, orientation = 'vertical')\n",
    "# Tick labels for x axis\n",
    "plt.xticks(x_values, feature_list, rotation='vertical')\n",
    "# Axis labels and title\n",
    "plt.ylabel('Importance'); plt.xlabel('Variable'); plt.title('Variable Importances');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## We had a model accuracy of 94% and found which variables were most predictive in predicting temperature!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}