Built a machine learning model to predict Spotify song popularity using audio features, metadata, and playlist metrics, achieving an R² of 0.835.
Also developed a content-based recommendation system that identifies songs with similar “vibes” using cosine similarity on engineered audio features.
Projects
Analyzed survival outcomes by sex, class, age + fare, transforming raw data using dplyr to perform statistical tests (chi-square, t-tests) and built a logistic regression model predicting survival, with 80% accuracy.
Analyzed NCAA KenPom data using Python, Pandas, Seaborn + Matplotlib to identify trends in tempo, efficiency, strength of schedule + win rates, leading to development of an Underrated Index highlighting underdog teams meriting a second look.