I have 10 independent variables to include in an OLS regression model. I know how to interpret the results and that for comparing models, the AICc value can be used, where a smaller value indicates a better model (ArcGIS Help).

But to find the best OLS model I cannot try all different combinations of the independent variables myself. So if I start with all 10 independent variables for the first model, how should I define, which independent variables to keep/remove to get a better model? Removing those variables where the coefficient is small? Removing those variables that are not statistically significant?

1 Answer 1


You should use "Exploratory Regression" tool which is a tool inside a complementary toolbox called "Supplementary Spatial Statistics". you can define many criteria for a passing model including R-square, VIF, p-values and ...

Download from: Supplementary Spatial Statistics Toolbox for ArcGIS 10


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.