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.
Together with the corresponding loop end node a boosting loop can be constructed. It repeatedly trains simple models and weights them according to their classification error. The algorithm used is AdaBoost.SAMME, i.e. is can also cope with multi-class problems. The first output contains the re- and over-sampled dataset, rows that have been predicted wrong are contained more often than correctly predicted rows.
0 | Any input data with nominal class labels |
0 | Possibly re-sampled training data, must be connected to the learner node inside the loop |
1 | Unaltered input data, must be connected to the predictor node inside the loop |