Class DataAudit.Config

java.lang.Object
edu.cmu.tetrad.data.audit.DataAudit.Config
Enclosing class:
DataAudit
public static final class DataAudit.Config extends Object
Thresholds for the audit's checks, with defaults. Instances are immutable; use the with-methods to derive variants. The defaults are rules of thumb, documented per field; they are deliberately conservative, in the sense that they flag situations most practitioners would want to know about.
Version:
$Id: $Id
Author:
josephramsey

  • Constructor Summary

    Constructors
    Constructor
    Description
    Config()
    Constructs a config with default thresholds.
    Config(int fewContinuousValues, int manyDiscreteLevels, int smallCellCount, double minExpectedPairwiseCell, double highCorrelation, double r2Determinism, double etaSquaredDeterminism, double adAlpha, int minAdSampleSize, double lowSampleRatio, double nearConstantFrequency, double nearConstantVariance)
    Constructs a config with the given thresholds.
    Config(int fewContinuousValues, int manyDiscreteLevels, int smallCellCount, double minExpectedPairwiseCell, double highCorrelation, double r2Determinism, double etaSquaredDeterminism, double adAlpha, int minAdSampleSize, double lowSampleRatio, double nearConstantFrequency, double nearConstantVariance, int serialMaxLag, double serialAlpha, double serialMinAbsAutocorrelation, int minSerialSampleSize, String serialGroupVariable)
    Constructs a config with the given thresholds, including the serial dependence settings.

  • Method Summary

    Modifier and Type
    Method
    Description
    DataAudit.Config
    withAdAlpha(double adAlpha)
    Returns a config identical to this one but with the given Anderson-Darling alpha.
    DataAudit.Config
    withBoundaryAtomMaxGapFraction(double boundaryAtomMaxGapFraction)
    Returns a config identical to this one but with the given small-gap fraction (see boundaryAtomMaxGapFraction).
    DataAudit.Config
    withBoundaryAtomMinMass(double boundaryAtomMinMass)
    Returns a config identical to this one but with the given minimum mass for BOUNDARY_ATOM.
    DataAudit.Config
    withCellDeterminismMaxCardinality(int cellDeterminismMaxCardinality)
    Returns a config identical to this one but with the given maximum distinct-value count for members of determining sets in the cell-determinism check.
    DataAudit.Config
    withCellDeterminismMaxSetSize(int cellDeterminismMaxSetSize)
    Returns a config identical to this one but with the given maximum determining-set size for the cell-determinism check (0 disables the check).
    DataAudit.Config
    withCellDeterminismTolerance(double cellDeterminismTolerance)
    Returns a config identical to this one but with the given relative tolerance for the cell-determinism check.
    DataAudit.Config
    withHighCorrelation(double highCorrelation)
    Returns a config identical to this one but with the given high-correlation threshold.
    DataAudit.Config
    withNonlinearDeterminismMaxSetSize(int nonlinearDeterminismMaxSetSize)
    Returns a config identical to this one but with the given nonlinear near-determinism maximum determining-set size (0 disables the check).
    DataAudit.Config
    withNonlinearR2Determinism(double nonlinearR2Determinism)
    Returns a config identical to this one but with the given nonlinear near-determinism threshold.
    DataAudit.Config
    withSentinelGapRatio(double sentinelGapRatio)
    Returns a config identical to this one but with the given minimum gap-to-median-spacing ratio for SENTINEL_VALUE.
    DataAudit.Config
    withSentinelMinCount(int sentinelMinCount)
    Returns a config identical to this one but with the given minimum cell count for SENTINEL_VALUE.
    DataAudit.Config
    withSentinelMinDistinct(int sentinelMinDistinct)
    Returns a config identical to this one but with the given minimum distinct-value count for the SENTINEL_VALUE check to run.
    DataAudit.Config
    withSentinelMinMass(double sentinelMinMass)
    Returns a config identical to this one but with the given minimum observed-value share for SENTINEL_VALUE.
    DataAudit.Config
    withSerialAlpha(double serialAlpha)
    Sets the serial correlation alpha value.
    DataAudit.Config
    withSerialGroupVariable(String serialGroupVariable)
    Returns a config identical to this one but with the given serial dependence grouping variable.
    DataAudit.Config
    withSmallCellCount(int smallCellCount)
    Returns a config identical to this one but with the given small-cell count threshold.

    Methods inherited from class Object

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

  • Constructor Details

    • Config

      public Config()
      Constructs a config with default thresholds.

    • Config

      public Config(int fewContinuousValues, int manyDiscreteLevels, int smallCellCount, double minExpectedPairwiseCell, double highCorrelation, double r2Determinism, double etaSquaredDeterminism, double adAlpha, int minAdSampleSize, double lowSampleRatio, double nearConstantFrequency, double nearConstantVariance)
      Constructs a config with the given thresholds. See the field documentation for meanings.
      Parameters:
      fewContinuousValues - threshold for CONTINUOUS_FEW_VALUES.
      manyDiscreteLevels - threshold for DISCRETE_MANY_LEVELS.
      smallCellCount - threshold for SMALL_MARGINAL_CELL.
      minExpectedPairwiseCell - threshold for SMALL_PAIRWISE_CELLS.
      highCorrelation - threshold for HIGH_CORRELATION.
      r2Determinism - threshold for NEAR_DETERMINISM_LINEAR.
      etaSquaredDeterminism - threshold for NEAR_DETERMINISM_DISCRETE_CONTINUOUS.
      adAlpha - alpha for NON_GAUSSIAN.
      minAdSampleSize - minimum column n for the Anderson-Darling test.
      lowSampleRatio - threshold for LOW_SAMPLE_RATIO.
      nearConstantFrequency - modal-frequency threshold for discrete NEAR_CONSTANT.
      nearConstantVariance - variance threshold for continuous NEAR_CONSTANT.

    • Config

      public Config(int fewContinuousValues, int manyDiscreteLevels, int smallCellCount, double minExpectedPairwiseCell, double highCorrelation, double r2Determinism, double etaSquaredDeterminism, double adAlpha, int minAdSampleSize, double lowSampleRatio, double nearConstantFrequency, double nearConstantVariance, int serialMaxLag, double serialAlpha, double serialMinAbsAutocorrelation, int minSerialSampleSize, String serialGroupVariable)
      Constructs a config with the given thresholds, including the serial dependence settings. See the field documentation for meanings.
      Parameters:
      fewContinuousValues - threshold for CONTINUOUS_FEW_VALUES.
      manyDiscreteLevels - threshold for DISCRETE_MANY_LEVELS.
      smallCellCount - threshold for SMALL_MARGINAL_CELL.
      minExpectedPairwiseCell - threshold for SMALL_PAIRWISE_CELLS.
      highCorrelation - threshold for HIGH_CORRELATION.
      r2Determinism - threshold for NEAR_DETERMINISM_LINEAR.
      etaSquaredDeterminism - threshold for NEAR_DETERMINISM_DISCRETE_CONTINUOUS.
      adAlpha - alpha for NON_GAUSSIAN.
      minAdSampleSize - minimum column n for the Anderson-Darling test.
      lowSampleRatio - threshold for LOW_SAMPLE_RATIO.
      nearConstantFrequency - modal-frequency threshold for discrete NEAR_CONSTANT.
      nearConstantVariance - variance threshold for continuous NEAR_CONSTANT.
      serialMaxLag - number of lags for SERIAL_DEPENDENCE (0 disables the check).
      serialAlpha - Ljung-Box alpha for SERIAL_DEPENDENCE.
      serialMinAbsAutocorrelation - minimum largest absolute autocorrelation for SERIAL_DEPENDENCE.
      minSerialSampleSize - minimum column n for the serial dependence check.
      serialGroupVariable - name of a discrete grouping variable for within-group autocorrelations, or null.

  • Method Details

    • withHighCorrelation

      public DataAudit.Config withHighCorrelation(double highCorrelation)
      Returns a config identical to this one but with the given high-correlation threshold.
      Parameters:
      highCorrelation - the new threshold.
      Returns:
      the new config.

    • withSmallCellCount

      public DataAudit.Config withSmallCellCount(int smallCellCount)
      Returns a config identical to this one but with the given small-cell count threshold.
      Parameters:
      smallCellCount - the new threshold.
      Returns:
      the new config.

    • withAdAlpha

      public DataAudit.Config withAdAlpha(double adAlpha)
      Returns a config identical to this one but with the given Anderson-Darling alpha.
      Parameters:
      adAlpha - the new alpha.
      Returns:
      the new config.

    • withSerialGroupVariable

      public DataAudit.Config withSerialGroupVariable(String serialGroupVariable)
      Returns a config identical to this one but with the given serial dependence grouping variable.
      Parameters:
      serialGroupVariable - the name of a discrete grouping variable, or null for no grouping.
      Returns:
      the new config.

    • withCellDeterminismMaxSetSize

      public DataAudit.Config withCellDeterminismMaxSetSize(int cellDeterminismMaxSetSize)
      Returns a config identical to this one but with the given maximum determining-set size for the cell-determinism check (0 disables the check).
      Parameters:
      cellDeterminismMaxSetSize - the new maximum set size.
      Returns:
      the new config.

    • withCellDeterminismMaxCardinality

      public DataAudit.Config withCellDeterminismMaxCardinality(int cellDeterminismMaxCardinality)
      Returns a config identical to this one but with the given maximum distinct-value count for members of determining sets in the cell-determinism check.
      Parameters:
      cellDeterminismMaxCardinality - the new maximum cardinality.
      Returns:
      the new config.

    • withCellDeterminismTolerance

      public DataAudit.Config withCellDeterminismTolerance(double cellDeterminismTolerance)
      Returns a config identical to this one but with the given relative tolerance for the cell-determinism check.
      Parameters:
      cellDeterminismTolerance - the new tolerance.
      Returns:
      the new config.

    • withNonlinearR2Determinism

      public DataAudit.Config withNonlinearR2Determinism(double nonlinearR2Determinism)
      Returns a config identical to this one but with the given nonlinear near-determinism threshold.
      Parameters:
      nonlinearR2Determinism - the new threshold.
      Returns:
      the new config.

    • withNonlinearDeterminismMaxSetSize

      public DataAudit.Config withNonlinearDeterminismMaxSetSize(int nonlinearDeterminismMaxSetSize)
      Returns a config identical to this one but with the given nonlinear near-determinism maximum determining-set size (0 disables the check).
      Parameters:
      nonlinearDeterminismMaxSetSize - the new maximum set size.
      Returns:
      the new config.

    • withSerialAlpha

      public DataAudit.Config withSerialAlpha(double serialAlpha)
      Sets the serial correlation alpha value.
      Parameters:
      serialAlpha - the serial correlation alpha value.
      Returns:
      a new Config object with the updated serial correlation alpha value.

    • withSentinelMinCount

      public DataAudit.Config withSentinelMinCount(int sentinelMinCount)
      Returns a config identical to this one but with the given minimum cell count for SENTINEL_VALUE.
      Parameters:
      sentinelMinCount - the new value.
      Returns:
      the new config.

    • withSentinelMinMass

      public DataAudit.Config withSentinelMinMass(double sentinelMinMass)
      Returns a config identical to this one but with the given minimum observed-value share for SENTINEL_VALUE.
      Parameters:
      sentinelMinMass - the new value.
      Returns:
      the new config.

    • withSentinelGapRatio

      public DataAudit.Config withSentinelGapRatio(double sentinelGapRatio)
      Returns a config identical to this one but with the given minimum gap-to-median-spacing ratio for SENTINEL_VALUE.
      Parameters:
      sentinelGapRatio - the new value.
      Returns:
      the new config.

    • withSentinelMinDistinct

      public DataAudit.Config withSentinelMinDistinct(int sentinelMinDistinct)
      Returns a config identical to this one but with the given minimum distinct-value count for the SENTINEL_VALUE check to run.
      Parameters:
      sentinelMinDistinct - the new value.
      Returns:
      the new config.

    • withBoundaryAtomMinMass

      public DataAudit.Config withBoundaryAtomMinMass(double boundaryAtomMinMass)
      Returns a config identical to this one but with the given minimum mass for BOUNDARY_ATOM.
      Parameters:
      boundaryAtomMinMass - the new threshold.
      Returns:
      the new config.

    • withBoundaryAtomMaxGapFraction

      public DataAudit.Config withBoundaryAtomMaxGapFraction(double boundaryAtomMaxGapFraction)
      Returns a config identical to this one but with the given small-gap fraction (see boundaryAtomMaxGapFraction).
      Parameters:
      boundaryAtomMaxGapFraction - the new fraction of the observed range.
      Returns:
      the new config.