since the R square contribution depends on the entering order of the variables in the model, could we try all different combinations and then get the average?
If we cannot, can any one provides a more generalized and widely acknowledged method?
Thanks very much!