Dropping one of the two highly co-related columns for Dimensionality Reduction

cor
dimensionality

#1

Dropping one of the highly correlated columns is one of the techniques used for dimensionality reduction.
How do we decide which column is to be dropped??


#2

@shuvayan - did you check for vif of those two variables by any chance? I remember dropping the variable which is having high vif among the two in one of the cases i worked.