Music Hidden Gems

Why this one is different from the others

The Steam and IMDb projects started with a dataset. This one did not. The data had to be built before any question could be asked of it.

Billboard chart history gives you which artists mattered and when. It says nothing about what people actually play. Last.fm gives you play counts and listener counts, but has no idea which artists ever charted.

The project is the join. Python collects the listening data one artist at a time from the Last.fm API, and SQL joins it to the chart history. Neither source can answer the question alone.

The measure

A hidden gem here is a track by a charting artist that never charted itself, and that the people who found it play repeatedly.

The second half is the part that took thought. Raw play count just finds popular songs again. What separates a gem is plays per listener, which is devotion rather than reach. A song with 190,000 listeners averaging 23 plays each is being lived with, not sampled.

What I checked

Last.fm listeners are not the general public. The platform's users skew younger, more online, and toward certain genres. Recent artists are over-represented against their chart-era peers. The measure describes the people who scrobble their listening, not everyone.

"Never charted" means never on the Hot 100. A song can be a well-known album track, or a hit in another country, and still qualify here. The claim is about the American singles chart specifically.

Artist name matching is the weak point of any join like this. Names differ between sources in punctuation, features and spelling. Every join on text has some loss, and this one is no exception.