Negative Data Fit Scores for Segment Models
Segment models and segment descriptions are computed using only input or "training" cases. Because of this, if there are "test" cases that contain information that was not present in the training cases, small values for the Recommendation Score or negative values for the Data Fit Score may be observed.
A Segment model that has a small value for the Recommendation Score or a negative value for the Data Fit Score indicates that the model may not make accurate predictions only on the test cases.
You may find the segments produced by the Segment model to be useful for other purposes and you can evaluate them using the Segment Viewer in Commerce Server Business Desk.
Ideally, the training cases and test cases will contain similar information.
If small values for the Recommendation Score or negative values for the Data Fit Score are observed, doing one of the following may alleviate the problem:
- Increase the value of the Sample Size****parameter.
- Decrease the value of the Measured Accuracy Sample Fraction parameter.
Be aware that increasing the value of the Sample Size parameter may result in longer time needed to build the Segment model.
Use caution in decreasing the value of the Measured Accuracy Sample Fraction parameter because this results in a smaller number of test cases. In general, it is better to have many test cases.
See Also
Viewing Analysis Model Configuration Tables
PredictorClient Object (Visual Basic)
Copyright © 2005 Microsoft Corporation.
All rights reserved.