adding factors to learning period

elena.vlad...@gmail.com

unread,

Mar 30, 2015, 3:59:21 AM3/30/15

to causal...@googlegroups.com

Hi!

How can I mark my dataset by flags of other intervention that could affect data on learning period?

Kay Brodersen

unread,

Mar 30, 2015, 4:03:30 AM3/30/15

to elena.vlad...@gmail.com, causal...@googlegroups.com

Hi Elena,

You could add a dummy variable to your set of regressors marking the training-period intervention with 1's. However, this assumes a fixed parametric form for the effect of the intervention. Another option would be to run CausalImpact twice, however this would produce overconfident inferences.

Kay

On 29 March 2015 at 13:32, <elena.vlad...@gmail.com> wrote:

Hi!
How can I mark my dataset by flags of other intervention that could affect data on learning period?

--
You received this message because you are subscribed to the Google Groups "CausalImpact" group.
To unsubscribe from this group and stop receiving emails from it, send an email to causalimpact...@googlegroups.com.
To post to this group, send email to causal...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/causalimpact/6acb1de7-9e32-4a9b-b88d-c253f5f99427%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Елена Серёгина

unread,

Mar 31, 2015, 10:14:53 AM3/31/15

to causal...@googlegroups.com, elena.vlad...@gmail.com

Kay, thank you, I'll try dummy vars!

понедельник, 30 марта 2015 г., 11:03:30 UTC+3 пользователь Kay Brodersen написал:

Reply all

Reply to author

Forward