A filter for adding the classification, the class distribution and an error flag to a dataset with a classifier. The classifier is either trained on the data itself or provided as serialized model.
Valid options are:
-D Turns on output of debugging information.
-W <classifier specification> Full class name of classifier to use, followed by scheme options. eg: "weka.classifiers.bayes.NaiveBayes -D" (default: weka.classifiers.rules.ZeroR)
-serialized <file> Instead of training a classifier on the data, one can also provide a serialized model and use that for tagging the data.
-classification Adds an attribute with the actual classification. (default: off)
-remove-old-class Removes the old class attribute. (default: off)
-distribution Adds attributes with the distribution for all classes (for numeric classes this will be identical to the attribute output with '-classification'). (default: off)
-error Adds an attribute indicating whether the classifier output a wrong classification (for numeric classes this is the numeric difference). (default: off)
@author fracpete (fracpete at waikato dot ac dot nz)
@version $Revision: 6900 $