Multicollinearity: "How to detect multicollinearity?
Formally, variance inflation factors (VIF) measure how much the variance of the estimated coefficients are increased over the case of no correlation among the X variables. If no two X variables are correlated, then all the VIFs will be 1.
If VIF for one of the variables is around or greater than 5, there is collinearity associated with that variable.
The easy solution is: If there are two or more variables that will have a VIF around or greater than 5, one of these variables must be removed from the regression model.
"
'via Blog this'
No comments:
Post a Comment