Best Subset Regression Options in GRM

These options are only available when the Best subsets option button is selected on the Quick Specs Dialog - Options tab or the GRM Analysis Wizard Extended Options - Advanced tab. As discussed in the Introductory Overview, the total number of all possible subsets (that need to be reviewed by STATISTICA) can become excessively large, when there are many effects in the model, and many large subset sizes are begin considered (via the Start and Stop options).

R squared

Select the R squared option button to use the whole model R-square (squared multiple regression) value as the criterion for choosing the best subset of predictor effects.

Adjusted R squared

Select the Adjusted R squared option button to use the adjusted model R-square (adjusted squared multiple regression) value as the criterion for choosing the best subset of predictor effects; the Adjusted R-squared is computed as 1 minus the ratio of the mean square residual over the mean square total.

Mallow's Cp

Select the Mallow's Cp option button to use Mallow's Cp value as the criterion for choosing the best subset of predictor effects. This measure of the quality of fit for a model tends to be less dependent (than the R-square) on the number of effects in the model, and hence, it tends to find the best subset that includes only the important predictors of the respective dependent variable.

Start, Stop

Enter values in the Start, Stop fields to determine the sizes of the subsets that will be considered during the search through all possible subsets. STATISTICA will begin the search with the subset size specified in the Start field, and will terminate the search after all subsets of the size specified in the Stop field have been evaluated.

Subsets to display

Enter a value in the Subsets to display field to determine the number of subsets of each size that will be displayed in the GRM Results dialog box in the Summary of best subset regression spreadsheet; for example, if you enter 10 in this field, you can later review (via option Summary of best subset regression on the GRM Results - Quick tab) the 10 best subsets (for each subset size, if the R squared option button is selected) according to the chosen criterion (see the R squared, Adjusted R squared, and Mallow's CP options above) for each size of subset that was considered (see options Start, Stop above).

Note: when you run Best subsets with multiple dependent variables, none of the univariate criteria are available (Mallow's CP, R squared, Adjusted R squared),and instead the selection of the best subset is based on the p for the multivariate test (Wilks' Lambda).

See also GRM - Index.