Assess Point-by-Point Models

In the point-by-point model view, you find controls and menu items specific to point-by-point models by using the Point-by-Point test plan. Use these tools to choose the best models:

To assess high-level model trends, use the Response Models tab at the test plan node. View the cross-section plots of all your response models at once. See Assess High-Level Model Trends.
In the point-by-point model node, you can assess all the alternative models for each operating point and decide which model type to choose for the selected operating point. Click in the alternative models list to view and compare the plots and statistics for each fit. See Assess Point-by-Point Fits Using Model Plots.
The alternative models list displays the value of your selection criteria, such as PRESS RMSE, for each model type, with the Best Model check box selected for the currently selected best model for the operating point. You select criteria when you create local multiple models. The toolbox automatically selects a best model for each operating point based on your selection criteria. Assess all the fits and, if desired, change the selected check box in the Best Model column.
Click Add Local Model in the Common Tasks pane to try adding one more model type. In the Model Setup dialog box, choose a model type to add. When you click OK, the toolbox fits the new model type to all operating points and then selects the model type as best if it is better (by your selection criteria) than any of the alternatives for an operating point. A dialog box informs you which operating points, if any, have a new best model.
Click Edit Model in the Common Tasks pane to change the list of alternative model types for every operating point. See Edit Point-by-Point Model Types.
Select Model > Summary Statistics to open the Summary Statistics dialog box. In this dialog box, select statistics to display in the alternative models list and in the Local summary statistics table. See Summary Statistics. If you are using validation data, the validation RMSE appears in the summary table for the operating point if there is validation data for the current operating point (global variables must match) for comparison with the model fit RMSE. See Using Validation Data.

Edit Point-by-Point Model Types

Click Edit Model in the Common Tasks pane to change the list of alternative model types for every operating point.

In the Point-by-Point Model Setup dialog box, choose model types to add or edit the existing model list. You can use any model available as one-stage models. You can choose the summary statistic to use as the selection criteria for deciding which model fits best to each operating point.

View the list of default point-by-point model types.
Click Add or Edit to add and change models.
When you add models, in the Model Setup dialog box, you can choose from all the global models available for a one-stage model with the same number of inputs as your current point-by-point model.
To choose a template to build a selection of models, click Template. Select from predefined templates for polynomials, radial basis functions, hybrid radial basis functions, and Gaussian process models. You can also save your own templates of any models you choose. See Create Alternative Models to Compare.
Select a statistic for selecting the best model in the Criteria list, such as RMSE or PRESS RMSE. The toolbox uses the statistic to select the best model type for each operating point. You can also change the choice of model for each operating point after the models are fitted.
Select statistics to display in the point-by-point model view by clicking Statistics.
To use data-based ranges for each operating point, leave the Automatic input ranges check box selected. Clear the check box only if you want to use instead the range for every local model that you set up in the test plan. Using data-based ranges for each operating point is helpful when the local input ranges for local operating points vary between operating points. This option is useful for diesel modeling as often there are a number of inputs at the local level, for example, main injection timing, pilot injection timing, rail pressure, boost pressure and EGR are common variables. The ranges of these variables vary over the global input space, such as torque and speed or fuel and speed. Adjusting the ranges for each operating point means that the inputs are scaled for modeling leading to better conditioned models.

After you add models to your list and close the dialog box, all the models you have chosen are fitted to each operating point individually, and the toolbox uses your selection criteria to select best fit to each operating point. Assess all the alternative models for each operating point.

Assess Point-by-Point Fits Using Model Plots

RMSE Plot

Use the RMSE plot to quickly identify problem operating points and navigate to an operating point of interest. Navigate to an operating point of interest by double-clicking a point in the plot to select the operating point in the other plots in the model view.

Response Surface Plot

This view shows the model surface in a variety of ways. The default view is a 3-D plot of the model surface.

You can choose which input factors to display by using the drop-down menus left of the plot. The unselected input factors are held constant, and you can change their values either by clicking the arrow buttons or by typing directly in the edit box on the left of the plot. Click Select Data Point to choose a point to plot.

Select the Plot list to switch to a Line, Contour, or Multiline plot.

Diagnostic Statistics Plot

The Diagnostic Statistics plot shows various scatter plots of statistics for assessing goodness-of-fit for the current model.

The statistics and factors available for plotting are model dependent. Choose the x- and y-axis factors using the drop-down menus.

Additional Plots

You can add or change plots by clicking the toolbar buttons, split buttons in plot title bars, or selecting an option from Current View in the context menu or View menu. The browser remembers your layout per test plan. You can add:

Predicted/Observed
Normal Plot
Validation Data
Model Definition

These plots are also used for one-stage models. See Assess One-Stage Models.

To view plots of the data for the current operating point, add Data Plots. Select View > Plot Variables to choose variables to plot. You can choose to view any of the data signals in the data set for the current operating point, including signals not being used in modeling. You can plot a pair of variables or plot a variable against record number. You can add more data plots.

You can also view values of input variables in the Operating Point pane.

Assess Point-by-Point Fits Using Statistics

For advice on using statistics such as PRESS RMSE to compare point-by-point models, see Compare Fits Using Statistics.