Courtesy: The Music Mom

This was a group project that I had worked on as part of my undergraduate thesis. The goal was to predict the emotion of a song as either Happy or Sad by analyzing its lyrics, and other features from its audio, such as pitch, tempo, and harmony. The data used was a subset of The Million Song Dataset. A combination of Machine Learning techniques (such as SVM, Random Forests, and Naive Bayes) and feature engineering was performed to build a classification model for this task. The abridged report, along with the entire thesis are viewable/downloadable below: