PBI is useful, but can be a royal pain to use. I think for this, I'll have a data pipeline something like...
Each week, grab the comments table for the most recent period. (sql)
Append it to the existing data. (python)
Complete my feature engineering and analysis (pbi)
Join it to any other relevant data (python)
Identify any new edge cases. (my brain)
Run the analytical compute. (probably python)
Update table, refresh dashboard, publish interesting bits on hive. (not sure on what tech i'll use)