An end-to-end classification project distinguishing the musical style of two 2nd-generation K-pop groups — Girls' Generation and 2NE1 — using audio features pulled from the Spotify API.
K-pop's global spread has been driven in large part by the idol groups that defined its earlier generations. This project asks a narrower question: what actually differentiates the music of one group from another, measured in audio features rather than reputation? I compared Girls' Generation and 2NE1 — two major acts from K-pop's 2nd generation — using track-level audio features (danceability, tempo, acousticness, liveness, valence, and more) pulled via the Spotify API.
Exploratory data analysis showed the two groups' tracks were broadly comparable across most features, with some tracks dominating specific dimensions — and a notable gap in valence, skewing higher for Girls' Generation. Before modeling, features were MinMax-scaled and categorical variables one-hot encoded. Several classifiers were trained and compared to predict artist from audio features alone, benchmarked against a proportional chance criterion (PCC) baseline.