Let’s attach the top two HDBSCAN labels and use them to exclude other recipes
DONE Figure out how to get the top two HDBSCAN labels
DONE Combine train and test datasets with HDBSCAN labels (likely with original recipe IDs and the NMF transformed dimensions)
TODO #MeaLeon For new data/searched recipes, perform NMF transformation, extract top two most probable labels, use to exclude stored recipes (data pipeline)
Derail
We talked about/looked at cars
stream_kyle redeemed MeaLeon for chilaquiles (Mexican)