Title: X-Validation with One-Class SVM
(Click on the image to get the full size)
With this example process, I intend to show how one might validate predictions done with a one class classifier. RapidMiner includes the One-Class SVM (part of libsvm) as introduced by Schoelkopf. I'm very interested in feedback concerning problems with or errors in this experiment.
Note that the data set (Sonar) is just a toy data set chosen for demonstration - learning a one class classifier on it won't give you any good results!
DESCRIPTION The basic idea is to partition the data set as usual and to train the one class classifier only on one class inside the cross validation, but to test it on both classes for the part that was left out for testing. So for training, we have to remove all examples with the label we don't want to train the classifier on (Filter). As the SVM operator expects the nominal label values to consist of only one value, we need to map the nominal value "Mine" to "Rock". When we apply the one class model to our test data, we get "inside/outside" as a prediction. These values have to be mapped back to the original corresponding nominal values "Rock" and "Mine". Afterwards, we can use the standard performance operator.
WARNING This process will only run with a newer version of RapidMiner, because the output of the One-Class SVM - inside/out for prediction(class) - recently has been changed. In earlier versions, one has to translate the resulting confidence values for the single prediction class back to two classes.
|Only Rock examples|
|Reduce to one class|
|Map Prediction to Rocks and Mines|
All versions of this Workflow are
CC0 1.0 Universal (Public Domain License)
Log in to add Tags
Shared with Groups (0)
Log in to add to one of your Packs
0.0 / 5
Log in to rate and see breakdown of ratings
In chronological order:
There are no workflows in myExperiment that use similar services to this Workflow.