NumericCleaner

Package

weka.filters.unsupervised.attribute

Synopsis

A filter that 'cleanses' the numeric data from values that are too small, too big or very close to a certain value (e.g., 0) and sets these values to a pre-defined default.

Options

The table below describes the options available for NumericCleaner.

Option

Description

attributeIndices

The selection of columns to use in the cleansing processs, first and last are valid indices.

closeTo

The number values are checked for whether they are too close to and get replaced by a default.

closeToDefault

The default value to replace values with that are too close.

closeToTolerance

The value below which values are considered close to.

debug

Turns on output of debugging information.

decimals

The number of decimals to round to, -1 means no rounding at all.

includeClass

If disabled, the class attribute will be always left out of the cleaning process.

invertSelection

If enabled the selection of the columns is inverted.

maxDefault

The default value to replace values that are above the maximum threshold.

maxThreshold

The maximum threshold above values are replaced by a default.

minDefault

The default value to replace values that are below the minimum threshold.

minThreshold

The minimum threshold below values are replaced by a default.

Capabilities

The table below describes the capabilites of NumericCleaner.

Capability

Supported

Class

Numeric class, Empty nominal class, No class, Relational class, Unary class, Missing class values, Date class, String class, Binary class, Nominal class

Attributes

Relational attributes, Missing values, Unary attributes, Nominal attributes, Numeric attributes, Empty nominal attributes, Date attributes, String attributes, Binary attributes

Min # of instances

0