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.

SVM Learner

This node trains a support vector machine on the input data. It supports a number of different kernels (HyperTangent, Polynomial and RBF). The SVM learner supports multiple class problems as well (by computing the hyperplane between each class and the rest), but note that this will increase the runtime.

The SVM learning algorithm used is described in the following papers: Fast Training of Support Vector Machines using Sequential Minimal Optimization, by John C. Platt and Improvements to Platt's SMO Algorithm for SVM Classifier Design, by S. S. Keerthi et. al.

If the optional PMML inport is connected and contains preprocessing operations in the TransformationDictionary those are added to the learned model.

Dialog Options

Class column
Choose the column that contains the nominal target variable.
Overlapping penalty
The overlapping penalty is useful in the case that the input data is not separable. It determines how much penalty is assigned to each point that is misclassified. A good value for it is 1.
Kernel type
There are a number of kernels to choose from. Each kernel has its own parameters, which appear in the configuration dialog just under the kernel.

Ports

Input Ports
0 Datatable with training data
1 Optional PMML port object containing preprocessing operations.
Output Ports
0 Trained Support Vector Machine

Views

SVM View
Shows the trained Support Vector Machines for each class with their corresponding support vectors.
This node is contained in KNIME Base Nodes provided by KNIME GmbH, Konstanz, Germany.