There are various methods for feature selection with the linear regression modeling. We have already been investigating possible solutions for this issue.
Currently, we can see the individual p-values and R-squared as well as R-squared matrix between the independent variables in the analysis. The most common approach to removing independent variables that are highly correlated with others is to do so one by one by removing the one with the highest R-squared in the multicollinearity matrix. As a secondary approach low value variables can be eliminated by evaluating the independent analysis.
This is currently a manual process; however we are considering options to add automated feature selection based on broad sweeping rules in a future version.
