I'm working on an ETA problem where I'm trying to estimate a time of arrival for a delivery. I have coordinates of pickup/destination, time of pick , infos about the rider, some other variables that are not very significant and we're tasked with esmating the time of arrival.
Will someone with experience on this particular field point out to me what's the most important features one could come up with that helped him/her achieve good results on such data?
Also using XGBoost regressor for this task, would a linear model work aswell? at least for ensembling purposes?