Publication date: 05/24/2021

Image shown hereValidation Column Role

The Validation Column role is available only in JMP Pro. For JMP, see Excluded Rows as Validation Holdback.

One way to create data partitions is to use the Validation Column role. The Validation Column role uses the column’s values to divide the data into parts. The column is assigned using the Validation role in the platform’s launch window. For information about how to create a validation column, see Make Validation Column.

Caution: The use of a validation column is platform specific. Different platforms use the levels of the validation column differently. See notes in Table A.1.

Table A.1 Validation Column by Platform

Platform

Train & Evaluate

Train & Tune

Train, Tune, & Evaluate

Notes

Fit Model

Fit Least Squares

Yes

No

No

If there are more than three levels, the validation column is ignored.

Stepwise Regression

No

Yes

Yes

If there are more than three levels, K-Fold Cross-Validation is used.

Logistic Regression

Yes

No

No

If there are more than three levels, the validation column is ignored.

Generalized Regression

No

Yes

Yes

If there are more than three levels, K-Fold Cross-Validation is used.

Partial Least Squares

No

Yes

Yes

If there are more than three levels, K-Fold Cross-Validation is used.

Predictive Models

Neural

No

Yes

Yes

If there are more than three levels, K-Fold Cross-Validation is used.

Partition

No

Yes

Yes

If there are more than three levels, the platform only uses rows with the three smallest values.

Bootstrap Forest

No

Yes

Yes

If there are more than three levels, the platform only uses rows with the three smallest values.

Boosted Tree

No

Yes

Yes

If there are more than three levels, the platform only uses rows with the three smallest values.

K Nearest Neighbors

No

Yes

Yes

If there are more than three levels, the platform only uses rows with the three smallest values.

Naive Bayes

No

Yes

Yes

If there are more than three levels, the platform only uses rows with the three smallest values.

Support Vector Machines

Yes

Yes

Yes

If there are more than three levels, K-Fold Cross-Validation is used.

Specialized Models

Functional Data Explorer

Yes

No

No

Must be created as a Grouped Random validation column.

If there are more than two levels, the smallest value defines the training set and all other values define the validation set.

Multivariate Models

Discriminant

Yes

Yes

Yes

If there are more than three levels, the platform only uses rows with the three smallest values.

Partial Least Squares

No

Yes

Yes

If there are more than three levels, K-Fold Cross-Validation is used.

Uplift

No

Yes

Yes

If there are more than three levels, the platform only uses rows with the three smallest values.

Want more information? Have questions? Get answers in the JMP User Community (community.jmp.com).