Hello, @lucianopaz and thanks a lot for this suggestion!
I am excited trying it, but can I please ask you to explain a couple of things?
I am hoping that by capturing the common movements of two cities, I would be able to get the latent demand for my company (which will help me get rid of confounders).
Running the code you suggested, I can see that correlation between variables indeed is captured ( I can see correlation matrix showing 0.8 at the posterior) However, the model does not get to predict the “external effects” I was originally asking about. Can you please say what I should be expecting from adding the MvNormal and correlations?
Also, can you please say if I am correct saying that sd_dist parameter should be related to my prior belief about the correlation between two variables? (Lower the variance of distribution => stronger belief about correlation)
Also, please say what you think about adding lag variables to the model with MvNormal?
(The ones @jessegrabowski suggested above)
Very much appreciated!