Shivani
← back to projects

Billboard “One-Hit Wonder”

Modeling chart reappearance with logistic regression

Role
Data + Modeling
Timeline
Jan 2025 – May 2025
Team
Solo
Tools
R
Python
Spotify API
Scikit-learn

Built a 2,000+ record dataset across 50+ years of charts and modeled reappearance at 95% accuracy.

Outcome · 95% accuracy

The question

Does a one-hit wonder ever come back? I wanted to know whether an artist’s single chart appearance could predict a future one — 50 years of Billboard data, tested.

The data

Built a 2,000+ record dataset combining Wikipedia, Spotify, and Last.fm with fuzzy matching across five decades of charts.

Modeled reappearance using logistic regression at 95% classification accuracy.