ORF: multiple treatments

Hi!

I have a question about implementing multiple treatments in ORF.

My idea is to estimate the effect of price and promotion on sales while controlling for high-dimensional covariates. In this case, I have two treatment variables: price and promotion. Price is a continuous variable; promotion can be either binary (the product is promoted or not) or continuous (the discount percentage). I wonder such a multiple treatments estimation can be done in ORF.

I search the example usage in ORF notebook (https://github.com/microsoft/EconML/blob/master/notebooks/Causal%20Forest%20and%20Orthogonal%20Random%20Forest%20Examples.ipynb). It seems that the multiple treatments example mentioned in this notebooks is not a "real" multiple treatments (See 3. Example Usage with Multiple Treatment Synthetic Data). It's more like a multiple-value discrete treatment, e.g., T has three discrete values 1, 2, 3 (but there is only one T).

But in DML notebook (https://github.com/microsoft/EconML/blob/master/notebooks/Causal%20Forest%20and%20Orthogonal%20Random%20Forest%20Examples.ipynb), I found a "true" multiple treatments case, in which the treatments include T and T^2 (See 3. Example Usage with Multiple Continuous Treatment Synthetic Data). This is close to what I want to implement, but not sure whether this is feasible in ORF.

And suggestions? Thanks!!

Start with the ORF notebook linked in the issue, especially its multiple-treatment synthetic-data example, and compare it with the DML notebook's multiple continuous treatment example. Determine whether separate price and promotion treatments are supported by ORF; done would be a documented answer or a clearly scoped implementation proposal.