Details on Classification Modeling – Options


This tool allows you to create classification models using the TIBCO Enterprise Runtime for R engine, without the need of writing any scripts yourself. A model page will be created (see The Model Page) and the model will be added to the Analytic Models panel.

  1. Open the Tools menu.

  2. Select Classification Modeling....

prd_classification_modeling_options_d.png

Top of Dialog

Option

Description

Name

The name of the model as you want it to be referenced in the Analytic Models panel.

Comment

A field for optional comments on the model.

Model method

Specifies the prediction model method. Choose from Logistic Regression and Classification Tree.

prd_question_mark_button.png

Displays more information about the currently selected model method.

Data table

Specifies the data table on which the model will be calculated.

Options Tab – Logistic Regression

The options available on this tab depend on which model method you have selected.

Option

Description

Use weights column

Select this option if you want to specify a weight column. A weight column is used to increase or decrease the importance of the values on specific rows by multiplication with the number in the weight column.

Maximum iterations

Specifies the maximum number of iterations to use in the regression calculation.

Convergence tolerance

Specifies the convergence tolerance of the model.

Options Tab – Classification Tree

Option

Description

Use weights column

Select this option if you want to specify a weight column. A weight column is used to increase or decrease the importance of the values on specific rows by multiplication with the number in the weight column.

Minimum split

Specifies the minimum number of observations in a node to consider splitting.

Complexity parameter

The complexity parameter is used for controlling the size of the classification tree and for selecting an optimal tree size. The building of the tree stops if the addition of another variable to the classification tree from the current node has a higher cost than the value of the complexity parameter. The building of the tree only continues if the overall lack of fit is decreased by a factor of the complexity parameter.

If the complexity parameter is set to zero then a tree will be built to its maximum depth, which may be very large.

Maximum depth

Specifies the maximum depth of any node in the tree.

Cross validation group size

Specifies the cross validation group size.

See also:

Details on Classification Modeling – General