AI Software Developers COMP-377

Lab Assignment #2 – Apply Logistic Regression and Support Vector

Machines algorithms to solve various classification problems.

Due Date: By the dropbox deadline.

Purpose: The purpose of this Lab assignment is to:

 Design and implement Python applications that incorporate Logistic
Regression solutions
 Design and implement Python applications that incorporate Support
Vector Machine (SVM) solutions

References: Read reference books, Stanford tutorial on logistic regression

(http://deeplearning.stanford.edu/tutorial/supervised/LogisticRegression/), and the
lecture slides. This material provides the necessary information that you need to
complete the exercises.

 Your submission MUST include the dataset files (if any) that you may have used. Instead of
using a dataset file, if you need call some function to fetch a dataset, then that should be
fine too.
 Your submission must be a zip file if compressed. Any other compression is not accepted.
 You MUST run the program of an exercise in the relevant .ipynb file and MUST retain the
output that gets generated (Note that the textual output of a program in an .ipynb file stays
in that .ipynb file if you do not delete the output).
 You MUST create a demo video of less than 3 minutes of your solution. Do not show
yourself in the demo video. Upload your video in your personal youtube account or google-
drive account and share its link with the instructor through the Comments box of submission
page as mentioned next (Do not share the video publicly).
 During submission at the dropbox, you should see a Comments box present near the bottom
of the submission page. Write the link of your video in this Comments box.
 Next, upload your solution and submit.

Exercise 1: Logistic Regression

Write a scikit-learn based application to predict the secondary school student performance using
a logistic regression model. The dataset is present in file student.cleaned.data.csv. The features
to be taken into account are traveltime, studytime, failures, famrel, freetime, gout, health.
The target should be G3. In G3 column, assume the values less than 10 to be 0, and the values
equal to or more than 10 to be 1. Evaluate the accuracy of the model.
(5 marks)

Lab #2 Page 1 of 3
AI Software Developers COMP-377

Exercise 2: Support Vector Machines

Write a scikit-learn based application to classify MNIST digits using a Support Vectors Machine
(SVM) model. The dataset is from http://yann.lecun.com/exdb/mnist/. You must use a
tensorflow function to just fetch the data. The description about this tensorflow function is in
this page: https://www.tensorflow.org/api_docs/python/tf/keras/datasets/mnist/load_data
Rest of the functionality must be accomplished using scikit-learn library. Train the model using
the top 60 rows out of 60000 rows of the training data (present in x_train; see below how to
obtain the training data in x_train). Test the model using top 10 rows out of 10000 rows of
test data (present in x_test; see below how to obtain the test data in x_test). Evaluate the
accuracy of the model.

Note: If you go down the aforementioned webpage you will see an example usage of the method
keras.datasets.mnist.load_data. It returns four objects of type ndarray. Those four
objects are x_train, x_test, y_train, y_test. The shape of x_train is (60000, 28,
28) implying 60000 rows of images, each image consists of 28 rows of pixels, 28 columns of
pixels. You need to reshape x_train to (60000, 784) so that it becomes a matrix of 60000
rows and 784 columns, thereby enabling it to be used by methods of sklearn models.
Similarly, the shape of x_test is (10000, 28, 28). You need to reshape x_test to (10000,
784) so that it can be used by sklearn methods. Note that the integer 784 = 28 * 28. (For an
analogy, you may recall that the sklearn method sklearn.datasets.load_digits returns an object
of type Bunch. The Bunch object in turn has an attribute named data that stores a ndarray or a
dataframe of shape (1797, 64). This shape implies that the ndarray object or the dataframe
object contains 1797 rows and 64 columns. Note that the integer 64 = 8 * 8.

(5 marks)

Evaluation:
Design and Functionality: 90%
Correct design and implementation of requirements
Code explanation if asked
Documentation of code using comments: 10%
At least a single-line comment for each functionality

Total 100%

Lab #2 Page 2 of 3
AI Software Developers COMP-377

You must name your Jupyter notebook file(s) according to the following rule:
YourFullname_COMP377Labnumber_Exercisenumber.ipynb

Example: JohnSmith_COMP377Lab1_Ex1.ipynb

Submission rules:

Submit your solution as a zip file that is named according to the following rule:
YourFullname_COMP377Labnumber.zip

Example: JohnSmith_COMP377Lab1.zip

Use 7-zip to compress files (https://www.7-zip.org/download.html).

Lab #2 Page 3 of 3