I have a question about weekly statistic forecast model.
In this model, can we control the difference of weeks in another year?
Suppose that, some food's demand will increase dramatically on 01/01(New Year).
・2014/01/01 - Week 1 FY14
・2015/01/01 - Week 52 FY14
In this case, I think we cannot control the Weekly Seasonality.
This case will happen in another situation(ex) Christmas, Halloween).
Do you want to normalize it or just calculate the seasonality within the same Year? Share your use case here with examples, will try our best to see if that can be achieved.
It seems that 1 Jan is on 1st week in every year.And you can set a flag on the specific week in this way.You can use these flags as dummy variables in your statistic model.
I hope it helps,
I have changed the end of fiscal year setting.
The week of 1 Jan moves, and It still works.
Hi Misbah.I just want to control the weekly seasonality within different year.
In this below case, the sales increase in 1/1-8, and decrease 1/9-16.
I think This is Seasonality.
But when we consider the weekly seasonality, can we catch it?
Week 1,2 are different in 2019 and 2020.
Hi Amaya.OK, I understood the case.But I think we can not control the another situation like public holidays.There are many public holidays in one year and they are in different week each year.Thanks for replay!