![]()
#Spotify playlist export to table how to#One really neat trick I had to learn for this report was how to implement cross-filter across dimensions. Some choices made while processing the raw data with Python really simplified the modelling in Power Query, namely unpivoting the audio features and keeping all the IDs in a single file (track_playlists.parquet). With the joins complete, the data modelling was also complete. #Spotify playlist export to table download#I joined this to the original tracks_playlists file (TrackPlaylists query if you download the report) on the track id, and from that point I had both audio features and all necessary values to join with the dimensions. The starting point of the fact was the file of audio features which contains the track id, the audio feature and the respective value. The fact was modelled by joining with all other queries to grab the dimension IDs from the automatic index column. For each query I kept only columns relevant to that dimension, removed duplicate rows and added an automatic index. The dimensions were created by duplicating the track playlists file into four different queries: playlists, albums, artists and tracks. Now let’s have a look at the data modelling in Power Query - though it was a light exercise to be honest. Audio Features FP was generated by Power BI when creating the audio features field parameters. The Audio Features Corr X and Y tables were used specifically for the correlation matrix visual. Note the correlation matrix always displays all the features. The audio features bar chart on the right uses the recently-released field parameters, controlled by the same-titled slicer on the left side.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |