Class IndTestConditionalGaussianLrt

java.lang.Object

edu.cmu.tetrad.search.test.IndTestConditionalGaussianLrt

All Implemented Interfaces:

IndependenceTest, RowsSettable, EffectiveSampleSizeSettable

public class IndTestConditionalGaussianLrt
extends Object
implements IndependenceTest, RowsSettable, EffectiveSampleSizeSettable

Performs a test of conditional independence X _||_ Y | Z1...Zn where all variables are either continuous or discrete. This test is valid for both ordinal and non-ordinal discrete variables.

Assumes a conditional Gaussian model and uses a likelihood ratio test. Drop-in replacement note: - setRows(...) now respects the provided argument (no random subsampling). - Row restriction is intersected with testwise deletion for the specific (x, y | z) variables. - Caches are cleared when row restrictions or discretization knobs change.

Effective sample size (added 2026-8-25). When an effective sample size nEff is set (see setEffectiveSampleSize(int)), the likelihood-ratio statistic 2 * (l1 - l0) is scaled by r = nEff / N, where N is the number of rows in the dataset, before being referred to the chi-square distribution. The degrees of freedom are unchanged. Since the log-likelihoods are row sums, this is the statistic that nEff independent rows would have produced. The value is relative to the full dataset: if setRows(List) restricts the test to m rows, or testwise deletion drops rows for a particular (x, y | z), the test behaves as if it had r * m independent rows, so a subsample or resample inherits the same per-row information content rather than the full nEff. (This differs from IndTestDegenerateGaussianLrt, which resets nEff to the actual row count on setRows, and from IndTestFisherZ, which keeps the absolute nEff regardless of setRows.) With nEff unset, r = 1 and the test is unchanged from the previous implementation.

Version:

$Id: $Id

Author:

josephramsey

Constructor Summary

Constructors

Constructor	Description
IndTestConditionalGaussianLrt(DataSet data, double alpha, boolean discretize)	Constructor.

Method Summary

Modifier and Type	Method	Description
IndependenceResult	checkIndependence(Node x, Node y, Set<Node> _z)	Returns an independence result that states whether x _||_ y | z and what the p-value of the test is.
boolean	determines(List<Node> z, Node y)	Determines whether a given list of nodes (z) determines a node (y).
double	getAlpha()	Returns the significance level of the independence test.
DataSet	getData()	Returns the data.
int	getEffectiveSampleSize()	Returns the effective sample size: the value set via setEffectiveSampleSize(int), or, if none has been set, the number of rows currently in use (the restriction from setRows(List) if any, otherwise the full dataset).
MissingValueSupport	getMissingValueSupport()	Declares what this test can do, natively, with data containing missing values.
double	getPValue()	Returns the probability associated with the most recently executed independence test, or Double.NaN if p value is not meaningful for this test.
List<Integer>	getRows()	Returns the user-specified row restriction (or null if none).
List<Node>	getVariables()	Returns the list of variables over which this independence checker is capable of determining independence relations.
IndependenceTest	indTestSubset(List<Node> vars)	This method returns an instance of the IndependenceTest interface that can test the independence of a subset of variables.
boolean	isVerbose()	Returns true iff verbose output should be printed.
void	setAlpha(double alpha)	Sets the significance level.
void	setEffectiveSampleSize(int nEff)	Sets the effective sample size, interpreted relative to the full dataset, or -1 (any negative value) to use the actual number of rows.
void	setMinSampleSizePerCell(int minSampleSizePerCell)	Sets the minimum sample size per cell for the independence test.
void	setNumCategoriesToDiscretize(int numCategoriesToDiscretize)	Sets the number of categories used to discretize variables.
void	setRows(List<Integer> rows)	Allows the user to set which rows are used in the test.
void	setVerbose(boolean verbose)	Sets whether verbose output should be printed.
String	toString()	Returns a string representation of this test.

Methods inherited from class Object

equals, getClass, hashCode, notify, notifyAll, wait, wait, wait

Methods inherited from interface IndependenceTest

canBeSubsampled, checkIndependence, determines, getCov, getDataSets, getSampleSize, getVariable, getVariableNames, isPValueAProbability

Constructor Details

IndTestConditionalGaussianLrt

public IndTestConditionalGaussianLrt(DataSet data,
 double alpha,
 boolean discretize)

Constructor.

Parameters:

data - The data to analyze.

alpha - The significance level.

discretize - Whether discrete children of continuous parents should be discretized.

Method Details

indTestSubset

public IndependenceTest indTestSubset(List<Node> vars)

This method returns an instance of the IndependenceTest interface that can test the independence of a subset of variables.

Specified by:

indTestSubset in interface IndependenceTest

Parameters:

vars - The sublist of variables to test for independence.

Returns:

An instance of the IndependenceTest interface.

See Also:

• IndependenceTest

checkIndependence

public IndependenceResult checkIndependence(Node x,
 Node y,
 Set<Node> _z)

Returns an independence result that states whether x _||_ y | z and what the p-value of the test is.

Specified by:

checkIndependence in interface IndependenceTest

Parameters:

x - a Node object

y - a Node object

_z - a Set object

Returns:

a IndependenceResult object

See Also:

• IndependenceResult

getPValue

public double getPValue()

Returns the probability associated with the most recently executed independence test, or Double.NaN if p value is not meaningful for this test.

Returns:

This p-value.

getVariables

public List<Node> getVariables()

Returns the list of variables over which this independence checker is capable of determining independence relations.

Specified by:

getVariables in interface IndependenceTest

Returns:

This list.

determines

public boolean determines(List<Node> z,
 Node y)

Determines whether a given list of nodes (z) determines a node (y).

Parameters:

z - The list of nodes to check if they determine y.

y - The node to check if it is determined by z.

Returns:

True if z determines y, false otherwise.

getAlpha

public double getAlpha()

Returns the significance level of the independence test.

Specified by:

getAlpha in interface IndependenceTest

Returns:

This level.

setAlpha

public void setAlpha(double alpha)

Sets the significance level.

Specified by:

setAlpha in interface IndependenceTest

Parameters:

alpha - This level.

getData

public DataSet getData()

Returns the data.

Specified by:

getData in interface IndependenceTest

Returns:

This data.

toString

public String toString()

Returns a string representation of this test.

Specified by:

toString in interface IndependenceTest

Overrides:

toString in class Object

Returns:

This string.

isVerbose

public boolean isVerbose()

Returns true iff verbose output should be printed.

Specified by:

isVerbose in interface IndependenceTest

Returns:

True if the case.

setVerbose

public void setVerbose(boolean verbose)

Sets whether verbose output should be printed.

Specified by:

setVerbose in interface IndependenceTest

Parameters:

verbose - True, if so.

setNumCategoriesToDiscretize

public void setNumCategoriesToDiscretize(int numCategoriesToDiscretize)

Sets the number of categories used to discretize variables.

Parameters:

numCategoriesToDiscretize - This number, by default 3.

setMinSampleSizePerCell

public void setMinSampleSizePerCell(int minSampleSizePerCell)

Sets the minimum sample size per cell for the independence test.

Parameters:

minSampleSizePerCell - The minimum sample size per cell.

getEffectiveSampleSize

public int getEffectiveSampleSize()

Returns the effective sample size: the value set via setEffectiveSampleSize(int), or, if none has been set, the number of rows currently in use (the restriction from setRows(List) if any, otherwise the full dataset).

Specified by:

getEffectiveSampleSize in interface EffectiveSampleSizeSettable

Returns:

the effective sample size

setEffectiveSampleSize

public void setEffectiveSampleSize(int nEff)

Sets the effective sample size, interpreted relative to the full dataset, or -1 (any negative value) to use the actual number of rows. A value of 0 is treated as unset. See the class Javadoc for the interaction with setRows and testwise deletion.

Configuration mutator: do not call concurrently with running tests.

Specified by:

setEffectiveSampleSize in interface EffectiveSampleSizeSettable

Parameters:

nEff - the effective sample size, or a nonpositive value to use the actual number of rows

getRows

public List<Integer> getRows()

Returns the user-specified row restriction (or null if none).

Specified by:

getRows in interface RowsSettable

Returns:

The current row restriction.

setRows

public void setRows(List<Integer> rows)

Allows the user to set which rows are used in the test. If null is passed, all rows are eligible (subject to testwise deletion for each test).

Specified by:

setRows in interface RowsSettable

Parameters:

rows - Row indices into the original DataSet.

getMissingValueSupport

public MissingValueSupport getMissingValueSupport()

Declares what this test can do, natively, with data containing missing values. The default is MissingValueSupport.NONE, meaning that missingness must be handled upstream (e.g., by imputation, or by supplying an EM-estimated covariance matrix if the test consumes one); tests that handle missing values themselves should override this.

Each likelihood-ratio calculation uses the rows complete on the variables involved in that calculation (the candidate rows are intersected with the rows having no missingness on those variables), i.e., test-wise deletion, unbiased only under MCAR.

Specified by:

getMissingValueSupport in interface IndependenceTest

Returns:

MissingValueSupport.TESTWISE.