Hey I was wondering how anyone here decided to deal with the scheme_management section. I would like to make things simpler by re-categorizing some of these values into new ones to make things simpler, but I tried looking up some of these values (like Waterboard and VWC and Trust) but there isn’t much clarity on how to think of them.
Hi,
On examination some of the values have a very small proportion of the total instances. - the None, SWC and Trust. So I regrouped these into the ‘other’ category. The idea was to reduce the factor levels so it does not skew the tree while running a random forest model.