Machine Learning Final Project

Overview

This repository contains the final project for the Machine Learning course, where we explore various machine learning algorithms, with a focus on Neural Networks for question 3, and Item Response Theory (IRT) Models with modifications for question 1. The IRT model has been enhanced by adding a new parameter to account for student effort, leading to improved performance in predictive tasks.

Key Files

• final_report.pdf: The main project report detailing the models and their performance.
• item_response.py: The original IRT model script.
• new_item_response.py: The modified IRT model that includes a new parameter αi to account for varying levels of student effort. This modification helped improve model accuracy and log-likelihood.
• neural_network.py: Implementation of the neural network for question 3.
• ensemble.py, knn.py, matrix_factorization.py, majority_vote.py: Other machine learning algorithms explored as part of the project.

How to Run the Code

1. Clone the repository:

   git clone https://github.com/VinayakMaharaj/CSC311-Final-Project
   

Install the necessary dependencies:

pip install -r requirements.txt

Run the Neural Network model for question 3:

python neural_network.py

To explore other models, use:

For k-NN:

python knn.py

For ensemble methods:

python ensemble.py

For matrix factorization:

python matrix_factorization.py

Results The results of the neural network and other models can be found in the generated output files and graphs, specifically:

final project q3 graph.png and final project q3 graph 2.png: Visualizations showing the model’s performance. final_report.pdf: Contains detailed analysis and discussion of the results.

Conclusion This project covers multiple machine learning algorithms, with a special focus on the neural network model for question 3 and the modified IRT model for question 1. The addition of the student effort parameter in the IRT model has provided a significant improvement in the accuracy and log-likelihood compared to the baseline model. Further improvements and experimentation with regularization and parameter tuning can enhance model performance even more.