I'm using vifcor
and vifstep
functions from the usdm package in R to detect multicollinearity. My understanding for vifcor
is that if I put the threshold as 0.9 for example it should give me all the variables with vif values <= 9. But the results showed much higher values (39, etc.).
So how exactly does it work and what is its relation to the vif value?
For vifcor
, th
is filtering based on the correlation threshold not VIF
. Use vifstep
if you want to filter using VIF
.