Fuzzy Rule Predictor

The first port contains the Fuzzy Rule Model that is applied to the test data contained in the second input port. The output data has then one additional column containing the predicted class attribute which is the best match for all rules.

Options

Unkown class handling

Select how to handle cases where the activation degree for all classes is below a certain threshold.

• Default: Use the minimum activation threshold from the learning algorithm.
• Ignore: If selected, no lower degree of class activation is set.
• Custom threshold: Instances where the activation lies below this threshold are classified as a missing (unknown) class. This is useful in cases where the feature space is not completely covered by rules.

Threshold

The threshold value between 0 and 1. Instances where the activation lies below this threshold are classified as unknown.

Custom prediction column name

Allows you to specify a customized name for the prediction column that is appended to the input table. If not checked, Prediction(target) (where target is the name of the target column of the provided regression model) is used as default.

Append columns with normalized class distribution

If selected, a column is appended for each class instance with the normalized probability of this row being a member of this class. The probability columns will have names like: P(targetColumn=value).

Probability column suffix

Suffix for the probability columns.

Views

This node has no views

Workflows

• 2_bisKNIME Forum
• Basketball Hubversion 2.2KNIME Hub
• Basketball Hubversion 2.5KNIME Hub
• Basketball Hubversion 2.5KNIME Hub
• Basketball Hubversion NiceKNIME Hub
• Show all 22 workflows

Links

• No links available

Developers