How to get forecast information for every draw from Causal impact

[Simha-Kalimipalli]

Hi There, Bayesian time series models typically return a 2d array of the forecast for every draw and every timestamp that is predicted. In the original documentation in the "7. Adjusting the model" section,(https://google.github.io/CausalImpact/CausalImpact.html#printing-a-summary-table) they outline a method to get this array from the model. Has that been implemented in the python package?

Thanks

[tcassou]

Hi @Simha-Kalimipalli! If I understood your question, you'd like to retrieve that prediction data on top of displaying graphs? If that's correct, have you tried this from the main README?

Basically the CausalImpact object holds a results attribute that contains the raw dataframe used for plots.

Let me know if I got your question wrong!

[Simha-Kalimipalli]

HI, there, I am referring to the the raw forecast (ie containing samples of the posterior distribution) that is typically available when using Bayesian sampling techniques. ie . I was wondering if one could input the number of samples that would be used for the prediction in Causal impact. This would be similar to Prophet's "Uncertainty in seasonality" function. https://facebook.github.io/prophet/docs/uncertainty_intervals.html#uncertainty-in-seasonality

Thanks,