I am working with the Python implementation of Google's CausalImpact package. My data is at daily frequency (365 observations per year); however, to inspect the effect of intervention, my pre-period is approximately 5 months long, and my post-period is approximately 5 months long. My full-range data (longer about 3 years) clearly shows yearly seasonality.
My question: how do I set the model up for yearly seasonality?
dated_data = df
pre_period = ['2018-01-01', '2018-06-05'] # Prior to known event
post_period = ['2018-06-06', '2019-11-25'] # After first, before second event
ci = CausalImpact(dated_data, pre_period, post_period,
prior_level_sd=None,
niter=1000,
nseasons=[{'period': ???}],
seasonal_duration=???)'
Any advice would be most welcomed.
