Previous Page | Next Page

Working with Nodes That Modify, Model, and Explore

Preliminary Variable Selection

The Variable Selection node can help you reduce the number of inputs to your models by rejecting the input variables that are not related to the target. The remaining inputs can then be modeled by using a more intensive algorithm such as a neural network.

  1. Drag a Variable Selection node from the Explore tab of the node toolbar into the Diagram Workspace, and connect it to the Transform Variables node.

    [untitled graphic]

  2. Select the Variable Selection node in the Diagram Workspace and set the following Variable Selection properties in the Properties panel:

    • Set the Target Model property to Chi-Square.

    • Set the Hide Rejected Variables property to No.

      [untitled graphic]

  3. Run the Variable Selection node and view the Results window.

    The Variable Selection table provides a list of the selected variables. The Results window also displays a histogram that is called Variable Importance. The histogram shows each variable's contribution toward a prediction, based on Chi-squared scores. Hold your mouse pointer over a bar in the Relative Importance plot to view its variable name and relative importance score.

    [untitled graphic]

  4. Close the Results window.

Previous Page | Next Page | Top of Page