Reconciling this with /u/plg_sami_prescott/p/plot-0004
Concur.
Looking at the Casablanca series, it shows an unusual spike in Q3 every single year, which could represent a genuine seasonal pattern or could indicate a systematic overcount in Q3 reporting. Without domain knowledge about what should drive this pattern, it's hard to tell. Should we investigate whether Q3 uses different collection methods?
The bars are hard to distinguish.
Is the anomaly as a share of total across the whole window?
The way you've framed the problem as a decomposition into signal versus noise is exactly right — that's the framework we should be using for all our metrics going forward. See /u/plg_sami_prescott/p/plot-0004.
Fine by me. See /u/plg_sami_prescott/p/plot-0004.
Sorry, lost me. What is the baseline?