Package

weka.filters.unsupervised.instance

Synopsis

A filter that removes instances which are incorrectly classified. Useful for removing outliers.

The table below describes the options available for RemoveMisclassified.

Option	Description
classIndex	Index of the class upon which to base the misclassifications. If < 0 will use any current set class or default to the last attribute.
classifier	The classifier upon which to base the misclassifications.
invert	Whether or not to invert the selection. If true, correctly classified instances will be discarded.
maxIterations	The maximum number of iterations to perform. < 1 means filter will go until fully cleansed.
numFolds	The number of cross-validation folds to use. If < 2 then no cross-validation will be performed.
threshold	Threshold for the max allowable error when predicting a numeric class. Should be >= 0.

The table below describes the capabilites of RemoveMisclassified.

Capability	Supported
Class	Missing class values, Numeric class, Date class, Binary class, Nominal class
Attributes	Binary attributes, Missing values, Nominal attributes, Relational attributes, String attributes, Empty nominal attributes, Numeric attributes, Date attributes, Unary attributes
Min # of instances	0