Package edu.cmu.tetrad.data

Class Discretizer

java.lang.Object

edu.cmu.tetrad.data.Discretizer

public class Discretizer
extends Object

Discretizes individual columns of discrete or continuous data. Continuous data is discretized by specifying a list of n - 1 cutoffs for n values in the discretized data, with optional string labels for these values. Discrete data is discretized by specifying a mapping from old value names to new value names, the idea being that old values may be merged.

Author:

josephramsey, Tyler Gibson

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	Discretizer.Discretization

Constructor Summary

Constructors

Constructor	Description
Discretizer(DataSet dataSet)	Constructs a new discretizer that discretizes every variable as binary, using evenly distributed values.
Discretizer(DataSet dataSet, Map<Node,DiscretizationSpec> specs)

Method Summary

Modifier and Type	Method	Description
DataSet	discretize()
static Discretizer.Discretization	discretize(double[] _data, double[] cutoffs, String variableName, List<String> categories)	Discretizes the continuous data in the given column using the specified cutoffs and category names.
void	equalCounts(Node node, int numCategories)	Sets the given node to discretized using evenly distributed values using the given number of categories.
void	equalIntervals(Node node, int numCategories)	Sets the given node to discretized using evenly spaced intervals using the given number of categories.
static double[]	getEqualFrequencyBreakPoints(double[] _data, int numberOfCategories)
void	setVariablesCopied(boolean unselectedVariabledCopied)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

Discretizer

public Discretizer(DataSet dataSet)

Constructs a new discretizer that discretizes every variable as binary, using evenly distributed values.

Discretizer

public Discretizer(DataSet dataSet,
 Map<Node,DiscretizationSpec> specs)

Method Details

getEqualFrequencyBreakPoints

public static double[] getEqualFrequencyBreakPoints(double[] _data,
 int numberOfCategories)

discretize

public static Discretizer.Discretization discretize(double[] _data,
 double[] cutoffs,
 String variableName,
 List<String> categories)

Discretizes the continuous data in the given column using the specified cutoffs and category names. The following scheme is used. If cutoffs[i - 1] < v <= cutoffs[i] (where cutoffs[-1] = negative infinity), then v is mapped to category i. If category names are supplied, the discrete column returned will use these category names.

Parameters:

cutoffs - The cutoffs used to discretize the data. Should have length c - 1, where c is the number of categories in the discretized data.

variableName - the name of the returned variable.

categories - An optional list of category names; may be null. If this is supplied, the discrete column returned will use these category names. If this is non-null, it must have length c, where c is the number of categories for the discretized data. If any category names are null, default category names will be used for those.

Returns:

The discretized column.

equalCounts

public void equalCounts(Node node,
 int numCategories)

Sets the given node to discretized using evenly distributed values using the given number of categories.

equalIntervals

public void equalIntervals(Node node,
 int numCategories)

Sets the given node to discretized using evenly spaced intervals using the given number of categories.

setVariablesCopied

public void setVariablesCopied(boolean unselectedVariabledCopied)

discretize

public DataSet discretize()

Returns:

- Discretized dataset.