Configuring the Rules block

In this section Rules block configuration is explained. The training of the Rules block requires the use of the offline trainer.

To configure the Rules block

Place a new Rules block from the Rules & Models palette on the blueprint.
The property page of an untrained rules block will denote that the model has not yet been trained.

The rules block is trained with the offline trainer block. Refer to the offline trainer help for further information.
The data type of the desired output target field must be string.
Note that only string fields in the input port of the offline trainer will be available for selection as Target field.

The properties page of a trained rules block will show information related to the training data and the rules that were found.
The Classes tab displays information of the data set and classification performance of the rules. The Name and number of Samples of the Classes in the data set can be seen. The Classification Matrix shows the classification accuracy of the rules. This example shows that of the 12932 samples that were Normal, 12802 was classified as Normal by the rules, 70 was classified as Low and 60 as High.
The Rules tab displays the rules identified to classify the data. Consider rule 1. It says that IF the FeedRate <= 46.5 AND MiscWater <= 17.83 THEN OverFlowDensity will be identified as belonging to class Low with a 54% accuracy. There were 14 cases during training that triggered this rule.