02-21-2018 10:05 AM
I have a dataset with 55 columns. I want to determine which of the columns can be used to determine the likely value of column A, they are a mix of character and numeric fields.
I am not sure which method to select/statistic to use. I have tried PROC CORR but this errors due to field type.
Any advice with this would be great.
02-22-2018 12:30 AM
Check PROC PLS, in its documentation, there is an example about variable's importance.
Or Check PROC HPGENSELECT,pick up the variable has the smallest P-value or the largest absolution parameter value.