Feature Selection

The Learning Wizard can perform feature selection in discrete node networks.

When there are many possible features available for a classification problem, we might want to select a subset of these features for the actual classification. We can learn such a subset by using mutual information.

We start with the target variable. Then we successively select features such that the next selcted feature has the highest sum of mutual information scores to the currently selected features.

Select target node

The target node for the features.

Number of features

The maximum number of nodes