Spotify provides a lot of valuable data to analyze. There's a bunch of data concerning tracks, artists, playlists and much more which can be analyzed at various angles. Based on two, already existing datasets I carried out vast analysis with elements of Feature Engineering using Python. You can find detailed descriptions and the whole analytical process in Spotify_EDA_and_FE.ipynb file uploaded into this repository.
Since datasets used for this analysis are too large to upload you can find data source links in .ipynb file.
You can also find find this notebook on Kaggle.