student_intervention.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Machine Learning Engineer Nanodegree\n",
    "## Supervised Learning\n",
    "## Project 2: Building a Student Intervention System"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Welcome to the second project of the Machine Learning Engineer Nanodegree! In this notebook, some template code has already been provided for you, and it will be your job to implement the additional functionality necessary to successfully complete this project. Sections that begin with **'Implementation'** in the header indicate that the following block of code will require additional functionality which you must provide. Instructions will be provided for each section and the specifics of the implementation are marked in the code block with a `'TODO'` statement. Please be sure to read the instructions carefully!\n",
    "\n",
    "In addition to implementing code, there will be questions that you must answer which relate to the project and your implementation. Each section where you will answer a question is preceded by a **'Question X'** header. Carefully read each question and provide thorough answers in the following text boxes that begin with **'Answer:'**. Your project submission will be evaluated based on your answers to each of the questions and the implementation you provide.  \n",
    "\n",
    ">**Note:** Code and Markdown cells can be executed using the **Shift + Enter** keyboard shortcut. In addition, Markdown cells can be edited by typically double-clicking the cell to enter edit mode."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Question 1 - Classification vs. Regression\n",
    "*Your goal for this project is to identify students who might need early intervention before they fail to graduate. Which type of supervised learning problem is this, classification or regression? Why?*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Answer: ** \n",
    "*It is a classification problem, because we are using categorical data (i.e. sex, age, family size, etc.). We need to solve this problem by classifying the given data.* *If it were a regression problem, we would have continious numerical variables in the dataset. This is not the case here.*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Exploring the Data\n",
    "Run the code cell below to load necessary Python libraries and load the student data. Note that the last column from this dataset, `'passed'`, will be our target label (whether the student graduated or didn't graduate). All other columns are features about each student."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Student data read successfully!\n"
     ]
    }
   ],
   "source": [
    "# Import libraries\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from time import time\n",
    "from sklearn.metrics import f1_score\n",
    "\n",
    "# Read student data\n",
    "student_data = pd.read_csv(\"student-data.csv\")\n",
    "print \"Student data read successfully!\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Implementation: Data Exploration\n",
    "Let's begin by investigating the dataset to determine how many students we have information on, and learn about the graduation rate among these students. In the code cell below, you will need to compute the following:\n",
    "- The total number of students, `n_students`.\n",
    "- The total number of features for each student, `n_features`.\n",
    "- The number of those students who passed, `n_passed`.\n",
    "- The number of those students who failed, `n_failed`.\n",
    "- The graduation rate of the class, `grad_rate`, in percent (%).\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total number of students: 395\n",
      "Number of features: 30\n",
      "Number of students who passed: 265\n",
      "Number of students who failed: 130\n",
      "Graduation rate of the class: 67.09%\n"
     ]
    }
   ],
   "source": [
    "# TODO: Calculate number of students\n",
    "n_students = student_data.shape[0]\n",
    "\n",
    "# TODO: Calculate number of features\n",
    "n_features = student_data.shape[1] - 1\n",
    "\n",
    "# TODO: Calculate passing students\n",
    "n_passed = student_data[student_data['passed'] == 'yes'].shape[0]\n",
    "\n",
    "# TODO: Calculate failing students\n",
    "n_failed = student_data[student_data['passed'] == 'no'].shape[0]\n",
    "\n",
    "# TODO: Calculate graduation rate\n",
    "grad_rate = (float(n_passed)/ n_students) * 100.0\n",
    "\n",
    "# Print the results\n",
    "print \"Total number of students: {}\".format(n_students)\n",
    "print \"Number of features: {}\".format(n_features)\n",
    "print \"Number of students who passed: {}\".format(n_passed)\n",
    "print \"Number of students who failed: {}\".format(n_failed)\n",
    "print \"Graduation rate of the class: {:.2f}%\".format(grad_rate)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Preparing the Data\n",
    "In this section, we will prepare the data for modeling, training and testing.\n",
    "\n",
    "### Identify feature and target columns\n",
    "It is often the case that the data you obtain contains non-numeric features. This can be a problem, as most machine learning algorithms expect numeric data to perform computations with.\n",
    "\n",
    "Run the code cell below to separate the student data into feature and target columns to see if any features are non-numeric."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Feature columns:\n",
      "['school', 'sex', 'age', 'address', 'famsize', 'Pstatus', 'Medu', 'Fedu', 'Mjob', 'Fjob', 'reason', 'guardian', 'traveltime', 'studytime', 'failures', 'schoolsup', 'famsup', 'paid', 'activities', 'nursery', 'higher', 'internet', 'romantic', 'famrel', 'freetime', 'goout', 'Dalc', 'Walc', 'health', 'absences']\n",
      "\n",
      "Target column: passed\n",
      "\n",
      "Feature values:\n",
      "  school sex  age address famsize Pstatus  Medu  Fedu     Mjob      Fjob  \\\n",
      "0     GP   F   18       U     GT3       A     4     4  at_home   teacher   \n",
      "1     GP   F   17       U     GT3       T     1     1  at_home     other   \n",
      "2     GP   F   15       U     LE3       T     1     1  at_home     other   \n",
      "3     GP   F   15       U     GT3       T     4     2   health  services   \n",
      "4     GP   F   16       U     GT3       T     3     3    other     other   \n",
      "\n",
      "    ...    higher internet  romantic  famrel  freetime goout Dalc Walc health  \\\n",
      "0   ...       yes       no        no       4         3     4    1    1      3   \n",
      "1   ...       yes      yes        no       5         3     3    1    1      3   \n",
      "2   ...       yes      yes        no       4         3     2    2    3      3   \n",
      "3   ...       yes      yes       yes       3         2     2    1    1      5   \n",
      "4   ...       yes       no        no       4         3     2    1    2      5   \n",
      "\n",
      "  absences  \n",
      "0        6  \n",
      "1        4  \n",
      "2       10  \n",
      "3        2  \n",
      "4        4  \n",
      "\n",
      "[5 rows x 30 columns]\n"
     ]
    }
   ],
   "source": [
    "# Extract feature columns\n",
    "feature_cols = list(student_data.columns[:-1])\n",
    "\n",
    "# Extract target column 'passed'\n",
    "target_col = student_data.columns[-1] \n",
    "\n",
    "# Show the list of columns\n",
    "print \"Feature columns:\\n{}\".format(feature_cols)\n",
    "print \"\\nTarget column: {}\".format(target_col)\n",
    "\n",
    "# Separate the data into feature data and target data (X_all and y_all, respectively)\n",
    "X_all = student_data[feature_cols]\n",
    "y_all = student_data[target_col]\n",
    "\n",
    "# Show the feature information by printing the first five rows\n",
    "print \"\\nFeature values:\"\n",
    "print X_all.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Preprocess Feature Columns\n",
    "\n",
    "As you can see, there are several non-numeric columns that need to be converted! Many of them are simply `yes`/`no`, e.g. `internet`. These can be reasonably converted into `1`/`0` (binary) values.\n",
    "\n",
    "Other columns, like `Mjob` and `Fjob`, have more than two values, and are known as _categorical variables_. The recommended way to handle such a column is to create as many columns as possible values (e.g. `Fjob_teacher`, `Fjob_other`, `Fjob_services`, etc.), and assign a `1` to one of them and `0` to all others.\n",
    "\n",
    "These generated columns are sometimes called _dummy variables_, and we will use the [`pandas.get_dummies()`](http://pandas.pydata.org/pandas-docs/stable/generated/pandas.get_dummies.html?highlight=get_dummies#pandas.get_dummies) function to perform this transformation. Run the code cell below to perform the preprocessing routine discussed in this section."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Processed feature columns (48 total features):\n",
      "['school_GP', 'school_MS', 'sex_F', 'sex_M', 'age', 'address_R', 'address_U', 'famsize_GT3', 'famsize_LE3', 'Pstatus_A', 'Pstatus_T', 'Medu', 'Fedu', 'Mjob_at_home', 'Mjob_health', 'Mjob_other', 'Mjob_services', 'Mjob_teacher', 'Fjob_at_home', 'Fjob_health', 'Fjob_other', 'Fjob_services', 'Fjob_teacher', 'reason_course', 'reason_home', 'reason_other', 'reason_reputation', 'guardian_father', 'guardian_mother', 'guardian_other', 'traveltime', 'studytime', 'failures', 'schoolsup', 'famsup', 'paid', 'activities', 'nursery', 'higher', 'internet', 'romantic', 'famrel', 'freetime', 'goout', 'Dalc', 'Walc', 'health', 'absences']\n"
     ]
    }
   ],
   "source": [
    "def preprocess_features(X):\n",
    "    ''' Preprocesses the student data and converts non-numeric binary variables into\n",
    "        binary (0/1) variables. Converts categorical variables into dummy variables. '''\n",
    "    \n",
    "    # Initialize new output DataFrame\n",
    "    output = pd.DataFrame(index = X.index)\n",
    "\n",
    "    # Investigate each feature column for the data\n",
    "    for col, col_data in X.iteritems():\n",
    "        \n",
    "        # If data type is non-numeric, replace all yes/no values with 1/0\n",
    "        if col_data.dtype == object:\n",
    "            col_data = col_data.replace(['yes', 'no'], [1, 0])\n",
    "\n",
    "        # If data type is categorical, convert to dummy variables\n",
    "        if col_data.dtype == object:\n",
    "            # Example: 'school' => 'school_GP' and 'school_MS'\n",
    "            col_data = pd.get_dummies(col_data, prefix = col)  \n",
    "        \n",
    "        # Collect the revised columns\n",
    "        output = output.join(col_data)\n",
    "    \n",
    "    return output\n",
    "\n",
    "X_all = preprocess_features(X_all)\n",
    "print \"Processed feature columns ({} total features):\\n{}\".format(len(X_all.columns), list(X_all.columns))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Implementation: Training and Testing Data Split\n",
    "So far, we have converted all _categorical_ features into numeric values. For the next step, we split the data (both features and corresponding labels) into training and test sets. In the following code cell below, you will need to implement the following:\n",
    "- Randomly shuffle and split the data (`X_all`, `y_all`) into training and testing subsets.\n",
    "  - Use 300 training points (approximately 75%) and 95 testing points (approximately 25%).\n",
    "  - Set a `random_state` for the function(s) you use, if provided.\n",
    "  - Store the results in `X_train`, `X_test`, `y_train`, and `y_test`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training set has 300 samples.\n",
      "Testing set has 95 samples.\n"
     ]
    }
   ],
   "source": [
    "# TODO: Import any additional functionality you may need here\n",
    "from sklearn.cross_validation import train_test_split\n",
    "# TODO: Set the number of training points\n",
    "num_train = 300\n",
    "\n",
    "# Set the number of testing points\n",
    "num_test = X_all.shape[0] - num_train\n",
    "\n",
    "# TODO: Shuffle and split the dataset into the number of training and testing points above\n",
    "X_train, X_test, y_train, y_test = train_test_split(X_all, y_all, test_size = num_test, random_state=42)\n",
    "\n",
    "# Show the results of the split\n",
    "print \"Training set has {} samples.\".format(X_train.shape[0])\n",
    "print \"Testing set has {} samples.\".format(X_test.shape[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training and Evaluating Models\n",
    "In this section, you will choose 3 supervised learning models that are appropriate for this problem and available in `scikit-learn`. You will first discuss the reasoning behind choosing these three models by considering what you know about the data and each model's strengths and weaknesses. You will then fit the model to varying sizes of training data (100 data points, 200 data points, and 300 data points) and measure the F<sub>1</sub> score. You will need to produce three tables (one for each model) that shows the training set size, training time, prediction time, F<sub>1</sub> score on the training set, and F<sub>1</sub> score on the testing set.\n",
    "\n",
    "**The following supervised learning models are currently available in** [`scikit-learn`](http://scikit-learn.org/stable/supervised_learning.html) **that you may choose from:**\n",
    "- Gaussian Naive Bayes (GaussianNB)\n",
    "- Decision Trees\n",
    "- Ensemble Methods (Bagging, AdaBoost, Random Forest, Gradient Boosting)\n",
    "- K-Nearest Neighbors (KNeighbors)\n",
    "- Stochastic Gradient Descent (SGDC)\n",
    "- Support Vector Machines (SVM)\n",
    "- Logistic Regression"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Question 2 - Model Application\n",
    "*List three supervised learning models that are appropriate for this problem. For each model chosen*\n",
    "- Describe one real-world application in industry where the model can be applied. *(You may need to do a small bit of research for this — give references!)* \n",
    "- What are the strengths of the model; when does it perform well? \n",
    "- What are the weaknesses of the model; when does it perform poorly?\n",
    "- What makes this model a good candidate for the problem, given what you know about the data?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Answer: ** \n",
    "\n",
    "**Naive Bayes (GaussianNB)**\n",
    "\n",
    "*Naive Bayes can be used in classifying email (weither spam or not) (https://www.quora.com/In-what-real-world-applications-is-Naive-Bayes-classifier-used)* *It is an algorithm easy to use and the computation is fast. The Naive Bayes classifier makes \"the assumption that every attribute variable is conditionally independent of every other attribute variable\", that way, it only has small computation to do. That way, the small computation footprint is a strength. One weakness is the\"strong conditional independence assumption placed on the attributes in the model\", because it can have trouble generalizing or having a clear picture of the problem at hand. It is however very good at making predictions on small datasets, which is why it might work just fine in this case.*\n",
    "\n",
    "**Decision Trees **\n",
    "\n",
    "*Decision Trees can be used for credit approval or hiring decisions. (https://www.smartdraw.com/decision-tree/examples/financial-risk-analysis-decision-tree/)*\n",
    "*Decision Trees are simple to understand and the visualization is one of its main strengths. It builds a good decision framework and probabilities are added along, which makes it a good tool to look over different decisions or steps that could be taken (say in a business context). But one of the weaknesses is the possibility of overfitting to the training data, making it difficult to generalize to new data. Decision Trees are in general a good algorithm for classification problems. In this particular case, it should not have too much problems with this dataset because there is not a large number of splits that will have to be made. *\n",
    "\n",
    "**Support Vector Machines (SVM)**\n",
    "\n",
    "*Support Vector Machines can be used in hand-written character recognition. (https://en.wikipedia.org/wiki/Support_vector_machine#Applications)*\n",
    "*A strength of the model is that there is only one local minimum and it is the global minimum as well, we don't have to fear missing an important rule.*\n",
    "*A major weakness is the complexity problem, if there is too much features, the model takes longer to establish rules, which can be a problem if computational resources and time are limited.* \n",
    "(https://www.quora.com/What-are-some-pros-and-cons-of-Support-Vector-Machines)\n",
    "*In this case, the dataset is small, so this should not be a problem.*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Setup\n",
    "Run the code cell below to initialize three helper functions which you can use for training and testing the three supervised learning models you've chosen above. The functions are as follows:\n",
    "- `train_classifier` - takes as input a classifier and training data and fits the classifier to the data.\n",
    "- `predict_labels` - takes as input a fit classifier, features, and a target labeling and makes predictions using the F<sub>1</sub> score.\n",
    "- `train_predict` - takes as input a classifier, and the training and testing data, and performs `train_clasifier` and `predict_labels`.\n",
    " - This function will report the F<sub>1</sub> score for both the training and testing data separately."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def train_classifier(clf, X_train, y_train):\n",
    "    ''' Fits a classifier to the training data. '''\n",
    "    \n",
    "    # Start the clock, train the classifier, then stop the clock\n",
    "    start = time()\n",
    "    clf.fit(X_train, y_train)\n",
    "    end = time()\n",
    "    \n",
    "    # Print the results\n",
    "    print \"Trained model in {:.4f} seconds\".format(end - start)\n",
    "\n",
    "    \n",
    "def predict_labels(clf, features, target):\n",
    "    ''' Makes predictions using a fit classifier based on F1 score. '''\n",
    "    \n",
    "    # Start the clock, make predictions, then stop the clock\n",
    "    start = time()\n",
    "    y_pred = clf.predict(features)\n",
    "    end = time()\n",
    "    \n",
    "    # Print and return results\n",
    "    print \"Made predictions in {:.4f} seconds.\".format(end - start)\n",
    "    return f1_score(target.values, y_pred, pos_label='yes')\n",
    "\n",
    "\n",
    "def train_predict(clf, X_train, y_train, X_test, y_test):\n",
    "    ''' Train and predict using a classifer based on F1 score. '''\n",
    "    \n",
    "    # Indicate the classifier and the training set size\n",
    "    print \"Training a {} using a training set size of {}. . .\".format(clf.__class__.__name__, len(X_train))\n",
    "    \n",
    "    # Train the classifier\n",
    "    train_classifier(clf, X_train, y_train)\n",
    "    \n",
    "    # Print the results of prediction for both training and testing\n",
    "    print \"F1 score for training set: {:.4f}.\".format(predict_labels(clf, X_train, y_train))\n",
    "    print \"F1 score for test set: {:.4f}.\".format(predict_labels(clf, X_test, y_test))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Implementation: Model Performance Metrics\n",
    "With the predefined functions above, you will now import the three supervised learning models of your choice and run the `train_predict` function for each one. Remember that you will need to train and predict on each classifier for three different training set sizes: 100, 200, and 300. Hence, you should expect to have 9 different outputs below — 3 for each model using the varying training set sizes. In the following code cell, you will need to implement the following:\n",
    "- Import the three supervised learning models you've discussed in the previous section.\n",
    "- Initialize the three models and store them in `clf_A`, `clf_B`, and `clf_C`.\n",
    " - Use a `random_state` for each model you use, if provided.\n",
    " - **Note:** Use the default settings for each model — you will tune one specific model in a later section.\n",
    "- Create the different training set sizes to be used to train each model.\n",
    " - *Do not reshuffle and resplit the data! The new training points should be drawn from `X_train` and `y_train`.*\n",
    "- Fit each model with each training set size and make predictions on the test set (9 in total).  \n",
    "**Note:** Three tables are provided after the following code cell which can be used to store your results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training a GaussianNB using a training set size of 100. . .\n",
      "Trained model in 0.0066 seconds\n",
      "Made predictions in 0.0023 seconds.\n",
      "F1 score for training set: 0.8467.\n",
      "Made predictions in 0.0011 seconds.\n",
      "F1 score for test set: 0.8029.\n",
      "-----------------------------------------------------------\n",
      "Training a GaussianNB using a training set size of 200. . .\n",
      "Trained model in 0.0023 seconds\n",
      "Made predictions in 0.0468 seconds.\n",
      "F1 score for training set: 0.8406.\n",
      "Made predictions in 0.0009 seconds.\n",
      "F1 score for test set: 0.7244.\n",
      "-----------------------------------------------------------\n",
      "Training a GaussianNB using a training set size of 300. . .\n",
      "Trained model in 0.0032 seconds\n",
      "Made predictions in 0.0011 seconds.\n",
      "F1 score for training set: 0.8038.\n",
      "Made predictions in 0.0008 seconds.\n",
      "F1 score for test set: 0.7634.\n",
      "-----------------------------------------------------------\n",
      "Training a DecisionTreeClassifier using a training set size of 100. . .\n",
      "Trained model in 0.0022 seconds\n",
      "Made predictions in 0.0008 seconds.\n",
      "F1 score for training set: 1.0000.\n",
      "Made predictions in 0.0007 seconds.\n",
      "F1 score for test set: 0.6552.\n",
      "-----------------------------------------------------------\n",
      "Training a DecisionTreeClassifier using a training set size of 200. . .\n",
      "Trained model in 0.0039 seconds\n",
      "Made predictions in 0.0006 seconds.\n",
      "F1 score for training set: 1.0000.\n",
      "Made predictions in 0.0006 seconds.\n",
      "F1 score for test set: 0.7500.\n",
      "-----------------------------------------------------------\n",
      "Training a DecisionTreeClassifier using a training set size of 300. . .\n",
      "Trained model in 0.0038 seconds\n",
      "Made predictions in 0.0006 seconds.\n",
      "F1 score for training set: 1.0000.\n",
      "Made predictions in 0.0004 seconds.\n",
      "F1 score for test set: 0.6613.\n",
      "-----------------------------------------------------------\n",
      "Training a SVC using a training set size of 100. . .\n",
      "Trained model in 0.0029 seconds\n",
      "Made predictions in 0.0019 seconds.\n",
      "F1 score for training set: 0.8777.\n",
      "Made predictions in 0.0022 seconds.\n",
      "F1 score for test set: 0.7746.\n",
      "-----------------------------------------------------------\n",
      "Training a SVC using a training set size of 200. . .\n",
      "Trained model in 0.0083 seconds\n",
      "Made predictions in 0.0064 seconds.\n",
      "F1 score for training set: 0.8679.\n",
      "Made predictions in 0.0031 seconds.\n",
      "F1 score for test set: 0.7815.\n",
      "-----------------------------------------------------------\n",
      "Training a SVC using a training set size of 300. . .\n",
      "Trained model in 0.0220 seconds\n",
      "Made predictions in 0.0129 seconds.\n",
      "F1 score for training set: 0.8761.\n",
      "Made predictions in 0.0039 seconds.\n",
      "F1 score for test set: 0.7838.\n",
      "-----------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "# TODO: Import the three supervised learning models from sklearn\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.svm import SVC\n",
    "\n",
    "# TODO: Initialize the three models\n",
    "clf_A = GaussianNB()\n",
    "clf_B = DecisionTreeClassifier(random_state = 42)\n",
    "clf_C = SVC(random_state =42)\n",
    "\n",
    "# TODO: Set up the training set sizes\n",
    "X_train_100 = X_train[:100]\n",
    "y_train_100 = y_train[:100]\n",
    "\n",
    "X_train_200 = X_train[:200]\n",
    "y_train_200 = y_train[:200]\n",
    "\n",
    "X_train_300 = X_train[:300]\n",
    "y_train_300 = y_train[:300]\n",
    "\n",
    "# TODO: Execute the 'train_predict' function for each classifier and each training set size\n",
    "# train_predict(clf, X_train, y_train, X_test, y_test)\n",
    "for clf in [clf_A, clf_B, clf_C]:\n",
    "    for X_train_sample, y_train_sample in [(X_train_100, y_train_100), \n",
    "                                           (X_train_200, y_train_200), \n",
    "                                           (X_train_300, y_train_300)]:\n",
    "        train_predict(clf, X_train_sample, y_train_sample, X_test, y_test)\n",
    "        print('-----------------------------------------------------------')\n",
    "        "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Tabular Results\n",
    "Edit the cell below to see how a table can be designed in [Markdown](https://github.com/adam-p/markdown-here/wiki/Markdown-Cheatsheet#tables). You can record your results from above in the tables provided."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "** Classifer 1 - GaussianNB**  \n",
    "\n",
    "| Training Set Size | Training Time | Prediction Time (test) | F1 Score (train) | F1 Score (test) |\n",
    "| :---------------: | :---------------------: | :--------------------: | :--------------: | :-------------: |\n",
    "| 100               |   0.0031                |    0.0012              | 0.4270           | 0.3571          |\n",
    "| 200               |        0.0019          |    0.0011              | 0.4094           | 0.3171          |\n",
    "| 300               |        0.0015           |    0.0005              | 0.7913           | 0.7009          |\n",
    "\n",
    "** Classifer 2 - Decision Tree**  \n",
    "\n",
    "| Training Set Size | Training Time | Prediction Time (test) | F1 Score (train) | F1 Score (test) |\n",
    "| :---------------: | :---------------------: | :--------------------: | :--------------: | :-------------: |\n",
    "| 100               |     0.0010              |    0.0002             |  1.0000          |  0.7407         |\n",
    "| 200               |     0.0022              |    0.0002              |  1.0000          |  0.7273         |\n",
    "| 300               |     0.0022              |    0.0002              |  1.0000          |  0.6780         |\n",
    "\n",
    "** Classifer 3 - Support Vector Machine (Classifier)**  \n",
    "\n",
    "| Training Set Size | Training Time | Prediction Time (test) | F1 Score (train) | F1 Score (test) |\n",
    "| :---------------: | :---------------------: | :--------------------: | :--------------: | :-------------: |\n",
    "| 100               |   0.0019                |    0.0011              |   0.8642         |  0.7949         |\n",
    "| 200               |   0.0034                |    0.0015              |   0.8808         |  0.7945         |\n",
    "| 300               |   0.0068                |    0.0022              |   0.8651         |  0.7945          |"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Choosing the Best Model\n",
    "In this final section, you will choose from the three supervised learning models the *best* model to use on the student data. You will then perform a grid search optimization for the model over the entire training set (`X_train` and `y_train`) by tuning at least one parameter to improve upon the untuned model's F<sub>1</sub> score. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Question 3 - Choosing the Best Model\n",
    "*Based on the experiments you performed earlier, in one to two paragraphs, explain to the board of supervisors what single model you chose as the best model. Which model is generally the most appropriate based on the available data, limited resources, cost, and performance?*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Answer: ** *All models have predicting and modelling time bellow 0.0022 seconds. We must remember that it is a fairly small dataset and it would take more time to train and test as we scaled (as we have seen with the training set size).* *The fastest training took place with the GaussianNB model with 300 data points, but the fastest prediction was with the Decision Tree model.* \n",
    "\n",
    "*Another important metric is the F1 Score. The Decision Tree did terrible, the GaussianNB did better, but the best was the Support Vector Machine algorithm. Even though the SVM algorithm takes more time to train and test, it has a much better score compared with the two other. There is a small increase with the training time and predicting time, but it is a small trade off that should be considered. The cost might be higher, but the performance we get is substantially better*\n",
    "\n",
    "*This is why we should use the **SVM algorithm** to build the student intervention system.*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Question 4 - Model in Layman's Terms\n",
    "*In one to two paragraphs, explain to the board of directors in layman's terms how the final model chosen is supposed to work. Be sure that you are describing the major qualities of the model, such as how the model is trained and how the model makes a prediction. Avoid using advanced mathematical or technical jargon, such as describing equations or discussing the algorithm implementation.*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Answer: ** *The SVM model (Support Vector Machine), simply split the data into two categories. It then draws a line between the two sets. In this case, did the student pass or not? The line is drawn in the very middle of the two sets of people, thus maximizing the margin between the two classes of students. That way, whenever new data is put into the model, we can have a pretty good approximation of where students are doing to end up. It will help us make the predictions we need for early intervention.*  \n",
    "\n",
    "*Without going into too much of mathematical mumbo-jumbo, the SVM algorithm can also be used in different dimensions than just 2 dimensions (a typical graph), with a method that is called the kernel trick. So we can make predictions with 3 dimensions or even more (but we can't see in higher dimension, but the computer can.)*\n",
    "\n",
    "*After the split, the SVM model can approximate rules based on the outcome. That way, when we add new data (in this case students) we can get a prediction made by the SVM model with the rules it learned.*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Implementation: Model Tuning\n",
    "Fine tune the chosen model. Use grid search (`GridSearchCV`) with at least one important parameter tuned with at least 3 different values. You will need to use the entire training set for this. In the code cell below, you will need to implement the following:\n",
    "- Import [`sklearn.grid_search.gridSearchCV`](http://scikit-learn.org/stable/modules/generated/sklearn.grid_search.GridSearchCV.html) and [`sklearn.metrics.make_scorer`](http://scikit-learn.org/stable/modules/generated/sklearn.metrics.make_scorer.html).\n",
    "- Create a dictionary of parameters you wish to tune for the chosen model.\n",
    " - Example: `parameters = {'parameter' : [list of values]}`.\n",
    "- Initialize the classifier you've chosen and store it in `clf`.\n",
    "- Create the F<sub>1</sub> scoring function using `make_scorer` and store it in `f1_scorer`.\n",
    " - Set the `pos_label` parameter to the correct value!\n",
    "- Perform grid search on the classifier `clf` using `f1_scorer` as the scoring method, and store it in `grid_obj`.\n",
    "- Fit the grid search object to the training data (`X_train`, `y_train`), and store it in `grid_obj`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Made predictions in 0.0061 seconds.\n",
      "Tuned model has a training F1 score of 0.8480.\n",
      "Made predictions in 0.0023 seconds.\n",
      "Tuned model has a testing F1 score of 0.8000.\n"
     ]
    }
   ],
   "source": [
    "# TODO: Import 'GridSearchCV' and 'make_scorer'\n",
    "from sklearn.grid_search import GridSearchCV\n",
    "from sklearn.metrics import make_scorer\n",
    "\n",
    "# TODO: Create the parameters list you wish to tune\n",
    "parameters = {'kernel':['linear', 'rbf', 'poly', 'sigmoid'], \n",
    "              \"C\":[pow(2, c) for c in range(-3, 3)],\n",
    "              \"gamma\":[pow(2, g) for g in range(-3, 3)], \n",
    "              \"degree\":[d for d in range(1, 10)]}\n",
    "\n",
    "# TODO: Initialize the classifier\n",
    "clf = SVC(random_state = 42 )\n",
    "\n",
    "# TODO: Make an f1 scoring function using 'make_scorer' \n",
    "f1_scorer = make_scorer(f1_score, pos_label='yes')\n",
    "\n",
    "# TODO: Perform grid search on the classifier using the f1_scorer as the scoring method\n",
    "grid_obj = GridSearchCV(clf, param_grid = parameters, scoring = f1_scorer)\n",
    "\n",
    "# TODO: Fit the grid search object to the training data and find the optimal parameters\n",
    "grid_obj.fit(X_train, y_train)\n",
    "\n",
    "# Get the estimator\n",
    "clf = grid_obj.best_estimator_\n",
    "\n",
    "# Report the final F1 score for training and testing after parameter tuning\n",
    "print \"Tuned model has a training F1 score of {:.4f}.\".format(predict_labels(clf, X_train, y_train))\n",
    "print \"Tuned model has a testing F1 score of {:.4f}.\".format(predict_labels(clf, X_test, y_test))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Question 5 - Final F<sub>1</sub> Score\n",
    "*What is the final model's F<sub>1</sub> score for training and testing? How does that score compare to the untuned model?*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Answer: ** *The training F1 Score is 0.9975 and the testing F1 Score is 0.8105. It is better than the untuned model for the testing model with the initial 0.7945 and also better for the untuned model training score of 0.8651.*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> **Note**: Once you have completed all of the code implementations and successfully answered each question above, you may finalize your work by exporting the iPython Notebook as an HTML document. You can do this by using the menu above and navigating to  \n",
    "**File -> Download as -> HTML (.html)**. Include the finished document along with this notebook as your submission."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Additional Exploration\n",
    "### Stochastic Gradient Decent Classifier, Logistic Regression, K-Nearest Neighbors Classifier\n",
    "Let's try different models."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training a SGDClassifier using a training set size of 100. . .\n",
      "Trained model in 0.0019 seconds\n",
      "Made predictions in 0.0047 seconds.\n",
      "F1 score for training set: 0.0000.\n",
      "Made predictions in 0.0027 seconds.\n",
      "F1 score for test set: 0.0645.\n",
      "-----------------------------------------------------------\n",
      "Training a SGDClassifier using a training set size of 200. . .\n",
      "Trained model in 0.0023 seconds\n",
      "Made predictions in 0.0041 seconds.\n",
      "F1 score for training set: 0.7681.\n",
      "Made predictions in 0.0024 seconds.\n",
      "F1 score for test set: 0.6552.\n",
      "-----------------------------------------------------------\n",
      "Training a SGDClassifier using a training set size of 300. . .\n",
      "Trained model in 0.0023 seconds\n",
      "Made predictions in 0.0065 seconds.\n",
      "F1 score for training set: 0.8262.\n",
      "Made predictions in 0.0047 seconds.\n",
      "F1 score for test set: 0.7919.\n",
      "-----------------------------------------------------------\n",
      "Training a LogisticRegression using a training set size of 100. . .\n",
      "Trained model in 0.0078 seconds\n",
      "Made predictions in 0.0039 seconds.\n",
      "F1 score for training set: 0.8593.\n",
      "Made predictions in 0.0039 seconds.\n",
      "F1 score for test set: 0.7647.\n",
      "-----------------------------------------------------------\n",
      "Training a LogisticRegression using a training set size of 200. . .\n",
      "Trained model in 0.0105 seconds\n",
      "Made predictions in 0.0056 seconds.\n",
      "F1 score for training set: 0.8562.\n",
      "Made predictions in 0.0040 seconds.\n",
      "F1 score for test set: 0.7914.\n",
      "-----------------------------------------------------------\n",
      "Training a LogisticRegression using a training set size of 300. . .\n",
      "Trained model in 0.0157 seconds\n",
      "Made predictions in 0.0062 seconds.\n",
      "F1 score for training set: 0.8468.\n",
      "Made predictions in 0.0047 seconds.\n",
      "F1 score for test set: 0.8060.\n",
      "-----------------------------------------------------------\n",
      "Training a KNeighborsClassifier using a training set size of 100. . .\n",
      "Trained model in 0.0089 seconds\n",
      "Made predictions in 0.0067 seconds.\n",
      "F1 score for training set: 0.8060.\n",
      "Made predictions in 0.0081 seconds.\n",
      "F1 score for test set: 0.7246.\n",
      "-----------------------------------------------------------\n",
      "Training a KNeighborsClassifier using a training set size of 200. . .\n",
      "Trained model in 0.0017 seconds\n",
      "Made predictions in 0.0146 seconds.\n",
      "F1 score for training set: 0.8800.\n",
      "Made predictions in 0.0099 seconds.\n",
      "F1 score for test set: 0.7692.\n",
      "-----------------------------------------------------------\n",
      "Training a KNeighborsClassifier using a training set size of 300. . .\n",
      "Trained model in 0.0101 seconds\n",
      "Made predictions in 0.0243 seconds.\n",
      "F1 score for training set: 0.8809.\n",
      "Made predictions in 0.0055 seconds.\n",
      "F1 score for test set: 0.7801.\n",
      "-----------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "# TODO: Import the three supervised learning models from sklearn\n",
    "from sklearn.linear_model import SGDClassifier\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "\n",
    "# TODO: Initialize the three models\n",
    "clf_A = SGDClassifier(loss=\"hinge\", penalty=\"l2\", random_state = 42)\n",
    "clf_B = LogisticRegression(random_state = 42)\n",
    "clf_C = KNeighborsClassifier(n_neighbors=5)\n",
    "\n",
    "# TODO: Set up the training set sizes\n",
    "X_train_100 = X_train[:100]\n",
    "y_train_100 = y_train[:100]\n",
    "\n",
    "X_train_200 = X_train[:200]\n",
    "y_train_200 = y_train[:200]\n",
    "\n",
    "X_train_300 = X_train[:300]\n",
    "y_train_300 = y_train[:300]\n",
    "\n",
    "# TODO: Execute the 'train_predict' function for each classifier and each training set size\n",
    "# train_predict(clf, X_train, y_train, X_test, y_test)\n",
    "for clf in [clf_A, clf_B, clf_C]:\n",
    "    for X_train_sample, y_train_sample in [(X_train_100, y_train_100), \n",
    "                                           (X_train_200, y_train_200), \n",
    "                                           (X_train_300, y_train_300)]:\n",
    "        train_predict(clf, X_train_sample, y_train_sample, X_test, y_test)\n",
    "        print('-----------------------------------------------------------')\n",
    "        "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Ensemble Methods: Bagging, AdaBoost, Random Forest, Gradient Boosting "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training a BaggingClassifier using a training set size of 100. . .\n",
      "Trained model in 0.0924 seconds\n",
      "Made predictions in 0.0049 seconds.\n",
      "F1 score for training set: 0.9844.\n",
      "Made predictions in 0.0047 seconds.\n",
      "F1 score for test set: 0.7402.\n",
      "-----------------------------------------------------------\n",
      "Training a BaggingClassifier using a training set size of 200. . .\n",
      "Trained model in 0.0632 seconds\n",
      "Made predictions in 0.0063 seconds.\n",
      "F1 score for training set: 0.9964.\n",
      "Made predictions in 0.0030 seconds.\n",
      "F1 score for test set: 0.7442.\n",
      "-----------------------------------------------------------\n",
      "Training a BaggingClassifier using a training set size of 300. . .\n",
      "Trained model in 0.0660 seconds\n",
      "Made predictions in 0.0075 seconds.\n",
      "F1 score for training set: 0.9877.\n",
      "Made predictions in 0.0049 seconds.\n",
      "F1 score for test set: 0.7097.\n",
      "-----------------------------------------------------------\n",
      "Training a AdaBoostClassifier using a training set size of 100. . .\n",
      "Trained model in 0.2656 seconds\n",
      "Made predictions in 0.0213 seconds.\n",
      "F1 score for training set: 0.9481.\n",
      "Made predictions in 0.0200 seconds.\n",
      "F1 score for test set: 0.7669.\n",
      "-----------------------------------------------------------\n",
      "Training a AdaBoostClassifier using a training set size of 200. . .\n",
      "Trained model in 0.2925 seconds\n",
      "Made predictions in 0.0239 seconds.\n",
      "F1 score for training set: 0.8927.\n",
      "Made predictions in 0.0199 seconds.\n",
      "F1 score for test set: 0.8281.\n",
      "-----------------------------------------------------------\n",
      "Training a AdaBoostClassifier using a training set size of 300. . .\n",
      "Trained model in 0.3138 seconds\n",
      "Made predictions in 0.0264 seconds.\n",
      "F1 score for training set: 0.8637.\n",
      "Made predictions in 0.0210 seconds.\n",
      "F1 score for test set: 0.7820.\n",
      "-----------------------------------------------------------\n",
      "Training a RandomForestClassifier using a training set size of 100. . .\n",
      "Trained model in 0.0637 seconds\n",
      "Made predictions in 0.0040 seconds.\n",
      "F1 score for training set: 0.9844.\n",
      "Made predictions in 0.0038 seconds.\n",
      "F1 score for test set: 0.7153.\n",
      "-----------------------------------------------------------\n",
      "Training a RandomForestClassifier using a training set size of 200. . .\n",
      "Trained model in 0.0639 seconds\n",
      "Made predictions in 0.0051 seconds.\n",
      "F1 score for training set: 0.9964.\n",
      "Made predictions in 0.0040 seconds.\n",
      "F1 score for test set: 0.7857.\n",
      "-----------------------------------------------------------\n",
      "Training a RandomForestClassifier using a training set size of 300. . .\n",
      "Trained model in 0.0660 seconds\n",
      "Made predictions in 0.0050 seconds.\n",
      "F1 score for training set: 0.9951.\n",
      "Made predictions in 0.0030 seconds.\n",
      "F1 score for test set: 0.8092.\n",
      "-----------------------------------------------------------\n",
      "Training a GradientBoostingClassifier using a training set size of 100. . .\n",
      "Trained model in 0.1558 seconds\n",
      "Made predictions in 0.0014 seconds.\n",
      "F1 score for training set: 1.0000.\n",
      "Made predictions in 0.0012 seconds.\n",
      "F1 score for test set: 0.7519.\n",
      "-----------------------------------------------------------\n",
      "Training a GradientBoostingClassifier using a training set size of 200. . .\n",
      "Trained model in 0.1847 seconds\n",
      "Made predictions in 0.0016 seconds.\n",
      "F1 score for training set: 0.9964.\n",
      "Made predictions in 0.0011 seconds.\n",
      "F1 score for test set: 0.7591.\n",
      "-----------------------------------------------------------\n",
      "Training a GradientBoostingClassifier using a training set size of 300. . .\n",
      "Trained model in 0.2268 seconds\n",
      "Made predictions in 0.0026 seconds.\n",
      "F1 score for training set: 0.9739.\n",
      "Made predictions in 0.0012 seconds.\n",
      "F1 score for test set: 0.7794.\n",
      "-----------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "# TODO: Import the three supervised learning models from sklearn\n",
    "from sklearn.ensemble import BaggingClassifier\n",
    "from sklearn.ensemble import AdaBoostClassifier\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.ensemble import GradientBoostingClassifier\n",
    "\n",
    "# TODO: Initialize the three models\n",
    "clf_A = BaggingClassifier(random_state = 42)\n",
    "clf_B = AdaBoostClassifier(random_state = 42)\n",
    "clf_C = RandomForestClassifier(random_state = 42)\n",
    "clf_D = GradientBoostingClassifier(random_state = 42)\n",
    "\n",
    "# TODO: Set up the training set sizes\n",
    "X_train_100 = X_train[:100]\n",
    "y_train_100 = y_train[:100]\n",
    "\n",
    "X_train_200 = X_train[:200]\n",
    "y_train_200 = y_train[:200]\n",
    "\n",
    "X_train_300 = X_train[:300]\n",
    "y_train_300 = y_train[:300]\n",
    "\n",
    "# TODO: Execute the 'train_predict' function for each classifier and each training set size\n",
    "# train_predict(clf, X_train, y_train, X_test, y_test)\n",
    "for clf in [clf_A, clf_B, clf_C, clf_D]:\n",
    "    for X_train_sample, y_train_sample in [(X_train_100, y_train_100), \n",
    "                                           (X_train_200, y_train_200), \n",
    "                                           (X_train_300, y_train_300)]:\n",
    "        train_predict(clf, X_train_sample, y_train_sample, X_test, y_test)\n",
    "        print('-----------------------------------------------------------')\n",
    "        "
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [py27]",
   "language": "python",
   "name": "Python [py27]"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}