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.

NG Learner (beta)

This node learns a classification model based on Neighborgrams. The model will cover all classes, for which Neighborgrams are constructed (as specified in the dialog). For details on the construction and individual dialog parameters, refer to the description of the NG Construct&View node. The few additional parameters are briefly commented on below:

Dialog Options

Purity
The purity value for the Neighborgram cluster candidates. The value must be in a range 0, 1 (though any value less than 0.6 is in most cases unreasonable). A higher value will cause more clusters to be identified.
Minimum coverage of a cluster
This parameter defines the termination criterion, i.e. the minimum of the accumulated coverage of each clusters. If the best next clusters covers less than this value, the clustering stops.

Ports

Input Ports
0 Training data.
Output Ports
0 Clustering statistics showing how many clusters were found in which universe (if any).
1 Classification model.

Views

Cluster Statistics
Shows simple cluster statistics (counts).
This node is contained in KNIME Neighborgrams & Parallel Universe Nodes provided by KNIME GmbH, Konstanz, Germany.