Hello
We are using the OneClassClassifier Node to train the one-class (OC) classifier.
We consider 4 different one-class classifiers as seen in the figure. But we are getting similar results to all those OC classifiers. That’s means that no matter OC we choose we get the same output as seen in the other figure.
Any help?
Could you provide a sample workflow maybe with dummy data. It is difficult to derive something from a screenshot.
Thanks @mlauber71
Yes this is the workflow and the iris.csv data
MultiOC_ForKNIME_forum.knwf (77.5 KB)
@malik I can verify this behavior using your workflow. I’m not sure what’s going on - I don’t have much experience with the Weka nodes myself - so I’ve asked internally for some assistance.
Hello @malik,
all the OneClassClassifier-nodes produce the same result because the densityOnly-option is set to ‘true’. In that case only the density estimate is used for the classification and no extra class probability obtained by the selected classifier is taken into account. More details can be found, e.g., in the respective publication: https://www.cs.waikato.ac.nz/~ml/publications/2008/hfw08-oneclassclassification.pdf (edited)
Best,
Martin
This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.
