I'm looking at energy consumption data for my thesis and I want to use a random forest model for predicting energy consumption using multiple features.
After I built, trained and validated my model, I want to interpret the data as much as possible.
Online I found that it is possible to look at feature importance to see which features were most influential in the outcome. Though my teacher asked me if there is a way to find out 'clusters of features which together lead to certain outcomes'. As an example he used (this example is totally made up):
A high income generally leads to more consumption of energy, being a man leads generally to less consumption of energy, but being a man and having a high income might lead to more consumption than being a woman with a high income
Is there any packages or functions that can find groups like this? So I guess instead of feature importance he is looking for something like 'group importance'? I'm not even sure anymore.