Predictive and Specialized Modeling > Model Screening > Model Screening Platform Options
Publication date: 05/24/2021

Image shown hereModel Screening Platform Options

Plot Actual by Predicted

(Available only for continuous responses.) Shows or hides a report of plots that overlay the actual by predicted points from all of the model fits. There are separate plots for the Training, Validation, and Test sets. The report contains a Methods legend with check boxes that enable you to specify which fits appear on the plots based on the method, fold, inner fold, and trial combination. Click the Deselect All button at the bottom of the legend to remove all model fits from the plots.

Notes:

If you select a subset of models from the Training, Validation, or Test reports and then select Plot Actual by Predicted, only the plots for the selected models are shown.

The actual by predicted plots are not supported by all of the modeling platforms. A list of unsupported platforms is shown below the plots.

Profiler

(Available only for continuous responses.) Shows or hides prediction profilers for each type of model fit. See Profiler in Profilers. If any type of folded crossvalidation is specified in the launch window, the selected model in the prediction profiler uses the fold, inner fold, and trial combination that corresponds to the final model. The final model is the model that produces the highest weighted average RSquare. The weight average RSquare is the average, weighted by number of observations, of the training RSquare, validation RSquare, and when a test set is used, the test RSquare.

Notes:

Confidence intervals are shown in prediction profilers for Standard Least Squares models and Generalized Regression models.

Prediction profilers are not available for K Nearest Neighbors models.

ROC Curve

(Available only for categorical responses.) Shows or hides a report of plots that overlay the ROC curves for all of the model fits. There are separate plots for the Training, Validation, and Test sets. The report contains two legends with check boxes that enable you to specify which fits appear on the plots. Click the Deselect All button at the bottom of the legend to remove all model fits from the plots. For more information about ROC Curves, see ROC Curve.

Notes:

If you select a subset of models from the Training, Validation, or Test reports and then select ROC Curves, only the plots for the selected models are shown.

The ROC plots are not supported by all of the modeling platforms. A list of unsupported platforms is shown below the plots.

Show Profit

(Available only for binary categorical responses that have a Profit Matrix column property.) Shows or hides the expected profit for each model.

Decision Threshold

(Available only for binary categorical responses.) Shows or hides Decision Thresholds reports for the training, validation, and test sets, if specified. Each report contains a graph of the distribution of fitted probabilities for each model, confusion matrices for each model, and classification graphs to adjust thresholds and compare the model fits. See Decision Thresholds Report.

Note: If you select a subset of models from the Training, Validation, or Test reports and then select Decision Threshold, only the information for the selected models is shown.

Save Results Table

Saves the information in the Validation report to a new data table titled Model Screening Statistics Validation Set. If there is a test set, such as when the Nested Crossvalidation option is specified, the information in the Test report is also saved to a new data table titled Model Screening Statistics Test Set.

Save KFold Results Table

(Available only if a Crossvalidation option is specified in the launch window.) Saves the information in the Summary Across the Folds report to a new data table titled KFold Results.

Save Prediction Formulas

(Available only if the live reports are included in the model report.) Saves the prediction formulas for the selected models to the data table. If you select a model in the Summary Across the Folds report and any type of fold crossvalidation is specified in the launch window, the prediction formula is for the fold, inner fold, and trial combination that corresponds to the final model. The final model is the model that produces the highest weighted average RSquare. The weight average RSquare is the average, weighted by number of observations, of the training RSquare, validation RSquare, and when a test set is used, the test RSquare.

Notes:

A prediction formula is not saved for Stepwise models.

A prediction formula is not saved for SVM models when the response is ordinal.

Optional Reports

Shows a submenu of additional report options.

Elapsed Time

Shows or hides a report that contains the total elapsed time that was spent fitting each type of model.

Cardinality of Predictors

Shows or hides a report of the number of levels and how many parameters are used in the linear model fit for each categorical predictor.

See Local Data Filters in Reports, Redo Menus in Reports, and Save Script Menus in Reports in Using JMP for more information about the following options:

Local Data Filter

Shows or hides the local data filter that enables you to filter the data used in a specific report.

Redo

Contains options that enable you to repeat or relaunch the analysis. In platforms that support the feature, the Automatic Recalc option immediately reflects the changes that you make to the data table in the corresponding report window.

Save Script

Contains options that enable you to save a script that reproduces the report to several destinations.

Save By-Group Script

Contains options that enable you to save a script that reproduces the platform report for all levels of a By variable to several destinations. Available only when a By variable is specified in the launch window.

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