keithkade/229-project

229-project

229-project's Issues

evaluation for graph approach

Baseline evaluation score

Spotify Playlist Scraping

Already wrote the example script for fetching, but need to extend that to fetch more and save them

Baseline NLP on lyrics

As a baseline, I'm thinking naive bayes to assign a sentiment score. Need to find a suitable training set.

Additional Lyrics Gathering

maybe necessary

Baseline Complete Pipeline

The algorithms can be stubs with fake results, but just want to connect all the pieces so we can input songs and get songs

Gather song lyrics

There are already some pretty big datasets that we can start with. Eventually we want a script that can fetch more.

Decide how to store the data efficiently

Independently enhance data with spotify/last.fm/better nlp

clean_data runs first, converting hdf to more manageable txts.

After that, we need two additional scripts, one to enhance (fill in missing year, add new features) with spotify/last.fm data, and one to enhance the nlp lyrics features.

playlists as a path through the graph

Extract features from timbre

Need to figure out how to utilize timbre

better regression

Baseline KNN

Baseline Classification Algo

Linear regression

better nlp

Merge different datasets

Need script to merge song features from the million song dataset, spotify data, and lyrics datasets.