There are legitimate reasons why stats wins out on interpretability. 1. Scaling is not hard 2. It is obvious to me "how to make these aggregations", but tha

There are legitimate reasons why stats wins out on interpretability. 1. Scaling is not hard 2. It is obvious to me "how to make these aggregations", but that is because I know statistics. Categories of variables treated as random effects can be interpreted both as a group (via variance parameters of the random effects) and individually via coefficients. 3. Bayesian estimation can even account for correlation of parameters and include it in the posterior prediction.