todo: seasonality procedure

[epogrebnyak]

As a simple approximation of the task - in a dataset I have some highly seasonal time sereis like these ones:

• http://minikep-db.herokuapp.com/all/series/INVESTMENT_bln_rub/m
• http://minikep-db.herokuapp.com/all/series/TRANSPORT_FREIGHT_bln_tkm/m

Can look up these in macrodash

As well as some time series where the seasonlaity is not apparent, but you also want to have it removed, as in GDP

Basically, I would like to store one or several seasonally adjusted versions for the indicator
and deliver it in API with sa parameter

http://minikep-db.herokuapp.com/all/series/TRANSPORT_FREIGHT_bln_tkm/m/sa (example, not working)

also sa can contain a verson like sa:x13, sa:something

a problem for me is that my code is python and most seasonal adjustment is in R
my own little snippet of R code

[epogrebnyak]

Some fun with R vs statmodels

[epogrebnyak]

Will move to a separate folder:

• https://github.com/mini-kep/user-charts/blob/master/seasonality.ipynb
• https://github.com/mini-kep/user-charts/blob/master/decomp.py