Add MAE loss function about sl3 HOT 2 OPEN

We're currently working on other developments but if you could point us to a manuscript on this -- as well as confirmation that the MAE satisfies the loss function properties outlined in, e.g., Dudoit and van der Laan (2005) -- we can get it on the to-do list. A PR would also be welcome, and we'd be happy to help in that process.

from sl3.

Hi @kartechbabu and @nhejazi, I believe the MAE identifies the conditional median of the outcome, given covariates.

@kartechbabu, could you please give us a little detail about your problem in terms of what you are trying to learn from your data? If you know the underlying prediction function that you're trying to estimate with SL, that would be really helpful for us to make sure the MAE is appropriate. If you're not sure, then we can help you get there based on the description of your problem.

from sl3.