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!
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.
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.
Answer:
We want to identify students who might need early intervention before they fail to graduate, so we have to seperate them into two classes based on whether they are likely to pass or fail. This is a classification problem as we are predicting discrete labels instead of continuous output.
In [1]:
# Import libraries
import numpy as np
import pandas as pd
from time import time
from sklearn.metrics import f1_score
# Read student data
student_data = pd.read_csv("student-data.csv")
print "Student data read successfully!"
In [2]:
student_data.head()
Out[2]:
In [51]:
student_data["passed"].value_counts()
Out[51]:
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_students.n_features.n_passed.n_failed.grad_rate, in percent (%).
In [3]:
# TODO: Calculate number of students
n_students = student_data.shape[0]
# TODO: Calculate number of features
n_features = student_data.shape[1] - 1
# TODO: Calculate passing students
n_passed = student_data["passed"].value_counts()["yes"]
# TODO: Calculate failing students
n_failed = student_data["passed"].value_counts()["no"]
# TODO: Calculate graduation rate
grad_rate = (265/395.0)*100
# Print the results
print "Total number of students: {}".format(n_students)
print "Number of features: {}".format(n_features)
print "Number of students who passed: {}".format(n_passed)
print "Number of students who failed: {}".format(n_failed)
print "Graduation rate of the class: {:.2f}%".format(grad_rate)
In this section, we will prepare the data for modeling, training and testing.
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.
Run the code cell below to separate the student data into feature and target columns to see if any features are non-numeric.
In [53]:
# Extract feature columns
feature_cols = list(student_data.columns[:-1])
# Extract target column 'passed'
target_col = student_data.columns[-1]
# Show the list of columns
print "Feature columns:\n{}".format(feature_cols)
print "\nTarget column: {}".format(target_col)
# Separate the data into feature data and target data (X_all and y_all, respectively)
X_all = student_data[feature_cols]
y_all = student_data[target_col]
# Show the feature information by printing the first five rows
print "\nFeature values:"
print X_all.head()
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.
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.
These generated columns are sometimes called dummy variables, and we will use the pandas.get_dummies() function to perform this transformation. Run the code cell below to perform the preprocessing routine discussed in this section.
In [54]:
def preprocess_features(X):
''' Preprocesses the student data and converts non-numeric binary variables into
binary (0/1) variables. Converts categorical variables into dummy variables. '''
# Initialize new output DataFrame
output = pd.DataFrame(index = X.index)
# Investigate each feature column for the data
for col, col_data in X.iteritems():
# If data type is non-numeric, replace all yes/no values with 1/0
if col_data.dtype == object:
col_data = col_data.replace(['yes', 'no'], [1, 0])
# If data type is categorical, convert to dummy variables
if col_data.dtype == object:
# Example: 'school' => 'school_GP' and 'school_MS'
col_data = pd.get_dummies(col_data, prefix = col)
# Collect the revised columns
output = output.join(col_data)
return output
X_all = preprocess_features(X_all)
print "Processed feature columns ({} total features):\n{}".format(len(X_all.columns), list(X_all.columns))
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:
X_all, y_all) into training and testing subsets.random_state for the function(s) you use, if provided.X_train, X_test, y_train, and y_test.
In [68]:
# TODO: Import any additional functionality you may need here
from sklearn.cross_validation import train_test_split
# TODO: Set the number of training points
num_train = 300
# Set the number of testing points
num_test = X_all.shape[0] - num_train
# TODO: Shuffle and split the dataset into the number of training and testing points above
X_train,X_test,y_train, y_test = train_test_split(X_all,y_all,test_size = num_test, random_state = 0)
# Show the results of the split
print "Training set has {} samples.".format(X_train.shape[0])
print "Testing set has {} samples.".format(X_test.shape[0])
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 F1 score. You will need to produce three tables (one for each model) that shows the training set size, training time, prediction time, F1 score on the training set, and F1 score on the testing set.
Answer:
The three supervised learning models that I've chosen are :
1. Decision Trees :
Decision Trees are widely used in several industries including medicine(to classify diseases based on patient's features for example), biomedical research, financial analysis(fraud detection,credit defaulting etc) for both classification and regression problem, astronomy to social media websites for predicting engagement/ad clicks for it's interpretability and versatality.
Weeknesses :
Reasons for choosing this model :
2. Support Vector Machine :
Support vector machines classify data by finding the maximum margin hyperplane that seperates class labels, it's also a very popular model like the other two, decision trees and K-nearest neighbors and used in industry for classification and regression tasks. Support Vector Machines have been successfully used on high dimensional data such as genetic data(protein structure prediction), music(song genre classification, music retrival), image classification(histogram based), image retrieval etc.
Strengths :
Weeknesses :
Reasons for choosing this model :
3. K Nearest Neighbor :
K nearest neighbor is a method of 'instance based learning'/lazy learning as the computation begins when we start predicting, it's also a non-parametric method. K nearest neighbor tries to find similar instances for each query and predicts based on their average/majority voting for classification and regression problem. It can be used in many different cases including content retrieval for photos, videos, text and recommending products etc. It's one of the most popular methods in data mining.
Strengths :
Weeknesses:
Reasons for choosing this model:
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:
train_classifier - takes as input a classifier and training data and fits the classifier to the data.predict_labels - takes as input a fit classifier, features, and a target labeling and makes predictions using the F1 score.train_predict - takes as input a classifier, and the training and testing data, and performs train_clasifier and predict_labels.
In [83]:
def train_classifier(clf, X_train, y_train):
''' Fits a classifier to the training data. '''
# Start the clock, train the classifier, then stop the clock
start = time()
clf.fit(X_train, y_train)
end = time()
# Print the results
print "Trained model in {:.4f} seconds".format(end - start)
def predict_labels(clf, features, target):
''' Makes predictions using a fit classifier based on F1 score. '''
# Start the clock, make predictions, then stop the clock
start = time()
y_pred = clf.predict(features)
end = time()
# Print and return results
print "Made predictions in {:.4f} seconds.".format(end - start)
return f1_score(target.values, y_pred, pos_label='yes')
def train_predict(clf, X_train, y_train, X_test, y_test):
''' Train and predict using a classifer based on F1 score. '''
# Indicate the classifier and the training set size
print "Training a {} using a training set size of {}. . .".format(clf.__class__.__name__, len(X_train))
# Train the classifier
train_classifier(clf, X_train, y_train)
# Print the results of prediction for both training and testing
print "F1 score for training set: {:.4f}.".format(predict_labels(clf, X_train, y_train))
print "F1 score for test set: {:.4f}.".format(predict_labels(clf, X_test, y_test))
print "\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:
clf_A, clf_B, and clf_C.random_state for each model you use, if provided.X_train and y_train.
In [84]:
# TODO: Import the three supervised learning models from sklearn
from sklearn.tree import DecisionTreeClassifier
from sklearn.svm import SVC
from sklearn.neighbors import KNeighborsClassifier
# TODO: Initialize the three models
clf_A = DecisionTreeClassifier(random_state =0)
clf_B = SVC(random_state = 0)
clf_C = KNeighborsClassifier()
training_sizes = [100,200,300]
# TODO: Execute the 'train_predict' function for each classifier and each training set size
# Decision Tree
for size in training_sizes:
train_predict(clf_A, X_train[:size], y_train[:size], X_test, y_test)
print "\n\n\n"
# Support Vector Machine
for size in training_sizes:
train_predict(clf_B, X_train[:size], y_train[:size], X_test, y_test)
print "\n\n\n"
# K Neareset Neighbor Classifier
for size in training_sizes:
train_predict(clf_C, X_train[:size], y_train[:size], X_test, y_test)
Edit the cell below to see how a table can be designed in Markdown. You can record your results from above in the tables provided.
Classifer 1 - DecisionTreeClassifier?
| Training Set Size | Training Time | Prediction Time (test) | F1 Score (train) | F1 Score (test) |
|---|---|---|---|---|
| 100 | 0.0050 seconds | 0.0010 seconds. | 1.0000 | 0.6942 |
| 200 | 0.0050 seconds | 0.0000 seconds | 1.0000 | 0.7132 |
| 300 | 0.0000 seconds | 0.0000 seconds | 1.0000 | 0.7167 |
Classifer 2 - SVM?
| Training Set Size | Training Time | Prediction Time (test) | F1 Score (train) | F1 Score (test) |
|---|---|---|---|---|
| 100 | 0.0000 seconds | 0.0000 seconds | 0.8591 | 0.7838 |
| 200 | 0.0180 seconds | 0.0150 seconds | 0.8693 | 0.7755 |
| 300 | 0.0310 seconds | 0.0000 seconds | 0.8692 | 0.7586 |
Classifer 3 - K-Nearest Neighbor
| Training Set Size | Training Time | Prediction Time (test) | F1 Score (train) | F1 Score (test) |
|---|---|---|---|---|
| 100 | 0.0000 seconds | 0.0000 seconds | 0.7972 | 0.7068 |
| 200 | 0.0000 seconds | 0.0070 seconds | 0.8571 | 0.7121 |
| 300 | 0.0030 seconds | 0.0000 seconds | 0.8722 | 0.7482 |
In [92]:
print y_train.value_counts()
print y_test.value_counts()
In [94]:
decision_tree_f1_average = (0.6942+0.7132+0.7167)/3.0
svm_f1_average = (0.7838 + 0.7755 + 0.7586)/3.0
k_nearest_f1_average = (0.7068+0.7121+0.7482)/3.0
print decision_tree_f1_average
print svm_f1_average
print k_nearest_f1_average
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 F1 score.
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?
Answer:
The model I would choose as the best model is SVM.
Reasons :
1. DecisionTreeClassifier shows clear signs of overfitting. It fits the training data perfectly with a F1-score of 1, but performs worse on the testing data compared to both SVM and k-nearest neighbor. So Decision Tree would clearly not be an appropriate model for this data set.
2. K-Nearest Neighbor actually shows quite stable performance over training and testing data sets and performs better both on the training and testing data sets steadily as the score increased with more training data(possibly because it found similar students with more training instances for the query instances). However, K-nearest's performance on the test data set is still poor compared to SVM.
3. SVM's average test score is 0.7726, beating both decision tree(average f1 on test set = 0.7080) and k-nearest neighbor(average f1 score 0.7223), based on scores SVM is the best choice. It's true that there's subtle differences of computation time for training and testing phases but for a small data set like this the differences are not that important.Answer:
The model that was chosen is called Support Vector Machine which is a linear seperator. Intuitively in the simplest case, we can imagine a 2D plane where we plot the data and labels on the x and y axis respective and we want to seperate the labels using a line. We can choose many lines for this task, assuming the labels are not overlapping, however, support vector machine will choose the "maximum margin" line, the line that has the biggest distance from the nearest points of both classes, i.e the line which is actually in the 'middle'. We choose this line to generalize the model to test data and avoid overfitting, a line too close to either of the classes can misclassify quickly.
For the higher dimension data sets instead of a line we map the datapoints to higher dimensions(with 'kernel trick') and find the maximum margin hyperplane to seperate the classes with as much gap as possible. For example in the image below a line could not have seperated circular data in 2D, so data has been mapped to 3D space where a clear seperating hyperplane was found, then the labels were used to classify the instances.
For practical purposes, choosing decision tree would have been more interpretable, but in this case would have led to overfitting (as we have seen in the table) and intervention of a student who's actually doing well because of a bad model would have led to negative consequences in this student's life. Choosing something like K-Nearest perhaps would have been stable perhaps, but not as interpretable as decision trees. However if we scale to millions of students, decision trees will also grow exponentially and K-Nearest neighbors woudld have to iterate over all the millions of students to find similar one's.
On the other hand, Support vector machine's clearly showed best performance so far and it's an widely used algorithm in the industry too, so SVM was chosen. Visualizing SVM is not as easy as decision tree's, but it has better performance.
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:
sklearn.grid_search.gridSearchCV and sklearn.metrics.make_scorer.parameters = {'parameter' : [list of values]}.clf.make_scorer and store it in f1_scorer.pos_label parameter to the correct value!clf using f1_scorer as the scoring method, and store it in grid_obj.X_train, y_train), and store it in grid_obj.
In [85]:
# TODO: Import 'GridSearchCV' and 'make_scorer'
from sklearn.grid_search import GridSearchCV
from sklearn.metrics import make_scorer
# TODO: Create the parameters list you wish to tune
parameters = {'kernel':('linear', 'poly','rbf'), 'C':[0.25,0.5,1, 10,50]}
# TODO: Initialize the classifier
clf = SVC()
# TODO: Make an f1 scoring function using 'make_scorer'
f1_scorer = make_scorer(f1_score,pos_label = "yes")
# TODO: Perform grid search on the classifier using the f1_scorer as the scoring method
grid_obj = GridSearchCV(clf,param_grid = parameters,scoring = f1_scorer)
# TODO: Fit the grid search object to the training data and find the optimal parameters
grid_obj.fit(X_train,y_train)
# Get the estimator
clf = grid_obj.best_estimator_
print clf.get_params()
# Report the final F1 score for training and testing after parameter tuning
print "Tuned model has a training F1 score of {:.4f}.".format(predict_labels(clf, X_train, y_train))
print "Tuned model has a testing F1 score of {:.4f}.".format(predict_labels(clf, X_test, y_test))
Answer:
Final models F1 Score for training : 0.8692 Final models F1 score for test : 0.7586.
It shows no difference from the 300 training point model chosen above, but it does not perform worse either. The most probable reason behind this situation is probably that grid search ended up choosing the default parameters despite given more options. I tried to reduce the C parameter, but it chose the value 1 again, which is also the default value, despite given more options for the Kernel it again chose the default version, which is "rbf" for non-linear datasets which is also optimum. The number of training points also don't vary as the total number of training points are 300.
So the grid search model is giving similar performance to the former one. This data set also is not balanced, there are more students who graduated than the one's who didn't graduate, perhaps that lead to more noise in the data which made SVM perform 0.7586 F1-score only in testing which is quite different from the training one.
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
File -> Download as -> HTML (.html). Include the finished document along with this notebook as your submission.