In traditional linear regression, we usually scale features before fitting the model. I suppose the same logic is valid for a bayesian GLMs right?
Thanks,
In traditional linear regression, we usually scale features before fitting the model. I suppose the same logic is valid for a bayesian GLMs right?
Thanks,
I say yes. You still have the similar kind of numerical problem even in Bayesian computation if the predictor matrix is badly conditioned. Preprocessing helps.