RegressionByDiscretization

Package

weka.classifiers.meta

Synopsis

A regression scheme that employs any classifier on a copy of the data that has the class attribute discretized. The predicted value is the expected value of the mean class value for each discretized interval (based on the predicted probabilities for each interval). This class now also supports conditional density estimation by building a univariate density estimator from the target values in the training data, weighted by the class probabilities.

For more information on this process, see

Eibe Frank, Remco R. Bouckaert: Conditional Density Estimation with Class Probability Estimators. In: First Asian Conference on Machine Learning, Berlin, 65-81, 2009.

Options

The table below describes the options available for RegressionByDiscretization.

Option	Description
classifier	The base classifier to be used.
debug	If set to true, classifier may output additional info to the console.
deleteEmptyBins	Whether to delete empty bins after discretization.
estimatorType	The density estimator to use.
minimizeAbsoluteError	Whether to minimize absolute error.
numBins	Number of bins for discretization.
useEqualFrequency	If set to true, equal-frequency binning will be used instead of equal-width binning.

Capabilities

The table below describes the capabilities of RegressionByDiscretization.

Capability	Supported
Class	Date class, Missing class values, Numeric class
Attributes	Numeric attributes, Nominal attributes, Empty nominal attributes, Date attributes, Binary attributes, Missing values, Unary attributes
Min # of instances	2