PREDICTING POPULARITY
// PROBLEM STATEMENT
In the streaming era, understanding what makes a song "popular" is a billion-dollar question. This project applies machine learning techniques to a massive dataset of Spotify tracks to identify the audio features most strongly correlated with commercial success.
// APPROACH
We utilized both regression (predicting exact popularity score) and classification (predicting "Hit" vs "Non-Hit") models. Extensive feature engineering was performed to extract insights from raw audio attributes like danceability, energy, valence, and tempo.
// KEY INSIGHTS
The analysis revealed that "loudness" and "energy" were strong predictors of popularity in the current pop landscape, while "acousticness" showed a negative correlation. The Random Forest classifier achieved the highest accuracy in predicting viral hits.