Course: Machine Learning (ECE-414)

This repository contains project work from a machine learning course taken at The Cooper Union.

The "mini-matlab" project work includes Bayesian estimation, conjugate priors, MLE, generative models, and logistic regression.

The "NMF_Recommender.py" file shows an implementation of Non-negative Matrix Factorization using Alternating Least Squares optimization. This implementation was used on a movie-ratings dataset from MovieLens to predict unknown user ratings for movies.

The "Using XGBoost" folder contains project work using the XGBoost algorithm for solving classification problems relevant to a Kaggle dataset (a fuller description of the datset can be found here: https://www.kaggle.com/miroslavsabo/young-people-survey). The Jupyter notebooks (represented as HTML files) demonstrate the results of the work done.

Our final project for this course was NBA game win-loss classification. We did the following: • Established a classification model for determining the outcome of NBA games • This involved webscraping, modeling team stats as a Gaussian process, using PCA to reduce the dimensionality of the feature space, as well as using linear regression, an SVM, and an autoencoder for classifier purposes. • Achieved an impressive 71% classification rate when tested on the 2015 NBA Season