0

In CausalImpact library in R, usually the pre period is considered as the period where the campaign was inactive and the post period is considered as the period after the intervention of the campaign. So the data for pre-period is considered as 0 and data for post period is considered as 1. Can we interchange this? Can we consider the pre period to be the period where the campaign is effective i.e, 1 and post period to be the period where the campaign is inactive i.e., 0?

T_S
  • 3
  • 2
  • I'm not familiar with CausalImpact (but looking at it now it looks super cool) though I am familiar with interrupted time series methods and assuming your campaign came first temporally speaking, it would generally be encoded as 0 and the inactive period would be encoded as 1. You can then interpret the post-period as the removal of a prior intervention, which is still a type of intervention. Were there issues with implementing it this way? – NicChr Aug 16 '23 at 11:49

0 Answers0