reichlab / 2017-2018-cdc-flu-contest Goto Github PK

In addition to LOSO fits training phase (everything but last 7 years), update fits for all component and ensemble models after each prospective-prediction year

Not sure if this is intentional, but pretty sure that the code at

returning a lot of "none" onsets... need to check.

some time after most backfill for the season has completed and before making predictions for that season, update data for the 2016-2017 season. when this is done, care should be exercised not to overwrite/replace data for preceeding seasons, which will have already been used to fit models.

e.g. get_onset_baseline() has this.

do this in the function that downloads and cleans data so that we have a standard name for the region

something went wrong in Region 6. Our models said week 2 was peak, CDC said week 52 was peak. baseline was 4.1. EW52 observation was 4.19, EW01 observation was 4.07, others were clearly above 4.1.

at least for prospective-prediction years, predictions only use data actually observed by prediction date