You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We can use the existing InterruptedTimeSeries class. But if we add in one (or more) control units as predictors then we essentially have Comparative Interrupted Time Series (CITS).
Highlight the similarities and differences to regular interrupted time series (no control units in predictors) and also synthetic control (no sum to 1 constraint, can have intercept).
To be clear, when predictors are not control units, then we have regular ITS. When we have one of more control units in the predictor list then we have CITS.
An acceptable PR would add another docs page in the concise style of our existing docs, but focused on the example in @juanitorduz 's blog post. It should be concise and to the point, but also approachable and clear. It should emphasise use of the existing InterruptedTimeSeries class and educate the reader on the similarities and differences between ITS, CITS, and SC.
Maybe we want to port that csv file into CausalPy and add it into the data folder. We would in fact have to update load_data in datasets.py and we should add an attribution in the docstring with a link back to Jaun's blog post URL.
The new docs page should be added into the existing Interrupted Time Series section in the docs.
The title of the notebook should be "Geo-Experimentation with Comparative Interrupted Time Series". The introduction should obviously make clear that the example is based on Juan's blog post.
