03-30-2018 12:46 AM
So I have 130 variables out of which around 70 are continuous and 60 are combination of categorical and demographics. Now to check multi-collinearity for continuous I am doing a correlation check but for the categorical I'm a little confused, should I go with Anova? If yes How can I do multiple variables at the same time in a single go?
Or there are any other ways to check multi-collinearity in categorical variables?
03-30-2018 06:46 AM
To tell you the truth, what I would do is the following:
I would just assume you have collinearity, and then use a modeling method that does a good job in the presence of collinearity and can handle categorical as well as continuous variables. That method is partial least squares regression (PROC PLS in SAS). Put all of your variables into the model (yes, all of them) and the ones that are predictive will have weights that are not close to zero, the ones that are not predictive will have weights close to zero.