Cross-Validation

Click the Test sample button on the Interactive Trees Specification dialog box - Validation tab to display the Cross-Validation dialog box. Use these options to specify a variable with a code to identify the observations that are to be included in the parameter estimation computations, i.e., in the fitting of the current model; the Analysis sample is comprised of these observations.

All observations that do not have the specified code value in the cross-validation variable will be excluded from the computations; however, you can compute predicted values and detailed residual statistics in the Results dialog box. The Cross-validation sample is comprised of the excluded observations have valid data values for all predictor variables and dependent variables. (Note that the Prediction sample is comprised of the observations that have valid data values for all predictor variables but missing data for one or more dependent variables in the current analysis.)

Sample Identifier Variable

Click the Sample Identifier Variable button to display a standard variable selection dialog. Select the variable containing the code value (as specified in the Code for analysis sample box, see below) to determine whether or not an observation belongs to the analysis sample.

Code for analysis sample

Type the code value (numeric or text) used in the Sample Identifier Variable in the Code for analysis sample box to identify observations belonging to the Analysis sample. You can double-click in this field (or press the F2 key on your keyboard) to review all unique code (text) values in the currently selected sample identifier variable.

Status

On/Off

Select either the On or Off option button under Status to turn the selection of observations based on the Sample identifier variable on or off.

OK

Click the OK button to close the dialog box after you are satisfied with the settings.

Cancel

Click the Cancel button to close the dialog box, disregarding any changes you have made.