[Back] [Up] [Next]

Scaling an input variable

Sometimes input variable data values are clustered in a certain range of the data. When this is the case, the points in the Scatter plot will bunch up in one or more dimensions. For more uniformly spaced points, you might try scaling one or more input variables. Each input variable can be scaled in the three common ways: Linear, Log, and Skew. To set the scaling method used for the input variable, click on the Linear, Log, or Skew radio buttons. If you choose the Skew scaling method then you should also enter a value greater than zero into the Skew factor text field to determine the function used for skew scaling.

Since the Scatter plot's input variables are likely to have wildly different data ranges, the Scatter plot provides an option to independently scale each input variable so it is in the range [0,1] so the resulting plot fits entirely in a cube. If you prefer to see the Scatter plot without this corrective scaling, you can turn off the Scale to cube check box on the Scatter plot attribute window's Appearance tab.