Hi there, is there anyway to do lead and lag on the features?

how would you predict using lead?

let's say I have a feature A that is available across training and test. I simply want to shift it forward by interval because there's a lag in target reaction. but I was just asking this generally - to take a lag or lead on the feature...

only lags are supported

what's the method?

what do you mean by method?

I mean the function to take the lag on the feature. I can see there's lag_transforms which is functioning on the target.

it's an internal function that takes a lag of all series, there's nothing special about it and it won't work with negative numbers if that's what you want to do

right I'm not trying to do that. the only reason I want that functionality is that I don't have to deal with the nans in the df. for example if I do df.shift I will get nans which I will further need to drop to train the model. I can see when I take lag on target by using

lag_transforms

or

lags

, they are filled by some values which I would assume it got handled.

they're not filled, they're dropped by default (

dropna=True

)