As written on this page :
Evaluation How Weka evaluates clustering depends on the cluster mode you select. Four different cluster modes are available (as buttons on the cluster mode panel):
Use training set
(default). After creating clustering, Weka classifies the training instances into clusters according to the presentation of the cluster and calculates the percentage of instances falling into each cluster. For example, the aforementioned clustering created by k-tool shows 43% (6 instances) in cluster 0 and 57% (8 instances) in cluster 1.- In the
Supplied test set
or Percentage split
Weka can evaluate clusters from separate test data if the cluster representation is probabilistic (for example, for EM). Classes to clusters evaluation
. In this mode, Weka first ignores the class attribute and generates clustering. Then, during the testing phase, he assigns classes to the clusters based on the value of most of the class attribute in each cluster. He then calculates a classification error based on this assignment and also shows the corresponding confusion matrix. An example of this for k-agents is shown below.
Sicco source share