Open
Description
Currently, extreme values at the edges of time series from partial windows can obscure the scale of "real" growth rates, making the output hard to use without postprocessing. Counterpoint: for the most recent data, maybe we'd want to try to get by with a partial window. Countercounterpoint: partial windows would have different meanings and behavior than the main growth rates with full windows, and the latest (& smallest) windows are unacceptably sensitive and would need to be filtered anyway.