The following node is available in the Open Source KNIME predictive analytics and data mining platform version 2.7.1. Discover over 1000 other nodes, as well as enterprise functionality at http://knime.com.

Boosting Learner Loop End

Together with the corresponding loop start node a boosting loop can be constructed. It repeatedly trains simple models and weighs them according to their classification error. The algorithm used is AdaBoost.SAMME, i.e. is can also cope with multi-class problems. The loop is stopped either after the maximum number of iterations has been reached or the weight for a model is only slightly above 0 (meaning the prediction error is too big).

Dialog Options

Real class column
The column from the second input table that contains the real class values for each row
Predicted class column
The column from the second input table that contains the predicted class values for each row
Number of iterations
The number of iterations the loop should be run i.e. the number of models to be learned

Ports

Input Ports
0 The trained model
1 The data with predicted classes and also the real class values
Output Ports
0 The boosted models together with their weights in data table
This node is contained in KNIME Ensemble Learning provided by KNIME GmbH, Konstanz, Germany.