Regression Model Tool

With the Regression Model Tool, you can create a regression model to display on scatter charts. To open the tool, click the edit icon, next to the 'Name' drop-down on the Regression Model tab.

You can use the context menu of a scatter chart to open the Scatter Chart Tool with the Regression Model tab active.

At the top of the tool you can use the toolbar to duplicate, rename, or delete your existing regression models.

Duplicates the currently selected item in the drop-down list (i.e., the active item).
Opens the Rename dialog, where a new name can be given to the active item.
Deletes the active item.

To create a regression model

  1. Select Create new from the Regression model drop-down list. Optionally, enter the name of regression model in the text field below. By default, a new regression model is named as 'Regression model <#>'. To modify the name of an existing regression model use the rename button .
  2. The following fields are populated as read-only:
    • Source chart  The name of the scatter chart (selected in the Scatter Chart Tool).
    • Source series  The name of the series for which you are creating the regression model.
    • X-Property type  The property type or log type of the source series (from the Data tab in the Scatter Chart Tool).
    • Y-Property type  The property type or log type of the source series (from the Data tab in the Scatter Chart Tool).
  3. From the Type drop-down list, select one of the following regression model types: Exponential, Linear, Logarithmic, Polynomial, or Power-Law. The tool displays the corresponding model parameters.
  4. Click the Autofit to Data button (recommended) to fill in the regression model parameters. You can also manually enter parameter values to better fit your data.
  5. Click Apply to create the regression model and keep the tool open, or click OK to create the regression model and close the tool. The regression model is immediately selected on the Scatter Chart Tool.
  6. Return to the Regression Model tab on the Scatter Chart Tool. To visualize the regression model on the scatter chart, click Apply to and keep the tool open, or click OK and close the tool.
    7. Important  If your scatter chart series is not visualized, ensure that the checkbox next to your chosen series is also checked in the Data tab.

You can view and adjust these parameters for regression models:

Type  Select the regression type from the drop-down list. You can select from one of the following models: Exponential, Linear, Logarithmic, Polynomial, or Power-Law.

Order  (Only for polynomial regression) You can add a polynomial order from a2, up to a6. Enter a value in the field or change the number with the arrow buttons.

Expression  Shows the mathematical equation used for the selected regression model.

a0-a6 parameters  (Only for polynomial regression) or a, b parameters (For all other regression types) Enter a value in the field for the regression parameters.

Cxunit/Cyunit  (Calculated for all regressions) Unit conversion factor constants that make the regression model compatible with the function tool. They are read-only and set to 1.0 in the display units of the X and Y quantities at the time the regression model is created. You can change the display units by right-mouse clicking on the unit label. When you change the display units, the Cxunit and Cyunit also change in order to keep intact the specified regression parameters. The constants keep the units consistent in the regression model so that the regression parameters are always dimensionless.

Weighting

The Unweighted and Weighted options control whether 'Autofit to Data' is performed without or with data weights. The Weighted option is only available for data from a 3D grid, where the cell volumes are used as weights, and well-log data, where the sample spacings are used as weights.

How to use the 'Autofit to Data' button

Click the Autofit to Data button to automatically populate the model parameters and calibrate them using the 'least squares regression'. This minimizes the vertical distance from the data points to the regression line using the smallest variance. The calculations are the same for each model type.