Multivariate Linear Regression with correlated input variables

Hi,
Thanks for the links. However, neither thread talks about correlated input variables. That’s the main part I would like to tackle.

How can one incorporate the uncertainties with this MvNormal construct?

obs = pm.MvNormal("obs", μ, chol=chol, observed=observed_df)

The observed argument here only accepts the X1 and X2 observed data and not their uncertainties.