Weights for an SVM model are automatically initialized to achieve the best average prediction across all target values. If you change a weight value, the percentage of correct predictions changes in the same way; for example, if you increase a weight value, the percent of correct predictions increases for the associated class.
For each target value, you can adjust the Weight column to indicate which class or classes should have the highest percent of correct predictions. The only practical way to adjust weights is to adjust them experimentally, that is, try several values until you get the desired results.
The weight value is normalized to be between 0 and 1 before the value is stored in the model.
Copyright © 2006, 2008, Oracle. All rights reserved.