When we look at data before these one off holidays (in April 2016 ) it seems to affect the data. This seems counter intuitive. That is , why is holiday data affecting data previous to their start date?
As well it seems to affect data in the future (june of 2024) a lot more than one would expect.
Should seasonality be used in conjunction with one off holidays to produce better results?
I am looking for some more details on one off holidays described here https://facebook.github.io/prophet/docs/handling_shocks.html#treating-covid-19-lockdowns-as-a-one-off-holidays
When we look at data before these one off holidays (in April 2016 ) it seems to affect the data. This seems counter intuitive. That is , why is holiday data affecting data previous to their start date?
As well it seems to affect data in the future (june of 2024) a lot more than one would expect.
Should seasonality be used in conjunction with one off holidays to produce better results?