I'm using this repo to store a few notes and code snippets to illustrate a project I'm working on at GA.
I want to know if topics or theme found in a broad genre of music, hiphop, are stronger in different regions of the US. Also, if any correlation exists between groups of artists by topic without region. The stretch goals I have planned include sentiment analysis, web UI, and additional visuals.
- Topic
- Artist
- Bio
- Hometown
- Affiliates
- Top Albums
- Similar artists
- Artists in topic
- Songs in topic
Last summer I scraped lyrics from a variety of web sources, general artist meta-data (Wikipedia), geolocation of artists, preprocessed quite a bit of data removing stop words.
A few people have asked about my wrapper for Gesim. I've moved it to it's own repo here: LDA Explorer