jdistlib.disttest

Class DistributionTest

java.lang.Object

jdistlib.disttest.DistributionTest

public class DistributionTest
extends java.lang.Object

Comparing two distributions

Author:

Roby Joehanes

Constructor Summary

Constructors

Constructor and Description
DistributionTest()

Method Summary

Modifier and Type	Method and Description
static double[]	ansari_bradley_test(double[] x, double[] y, boolean force_exact) Return the two-sided test of Ansari-Bradley.
static double[]	ansari_bradley_test(double[] x, double[] y, boolean force_exact, TestKind kind) Ansari-Bradley test.
static double[]	bartlett_test(double[] x, int[] group) Bartlett's test
static double[]	binomial_test(int n_success, int n, double p, TestKind kind) Binomial test
static double[]	diptest_presorted(double[] x) Perform Hartigan's dip test, assuming the minimum test statistics D is zero.
static double[]	diptest(double[] x) Perform Hartigan's dip test, assuming the minimum test statistics D is zero.
static double[]	fligner_test(double[] x, int[] group) Fligner-Killeen test
static double[]	kolmogorov_smirnov_test(double[] X, double[] Y) Compute the Kolmogorov-Smirnov test to test between two distribution, two-sided, exact p-value.
static double[]	kolmogorov_smirnov_test(double[] X, double[] Y, boolean isExact) Compute the Kolmogorov-Smirnov test to test between two distribution, two-sided.
static double[]	kolmogorov_smirnov_test(double[] X, double[] Y, TestKind kind) Compute the Kolmogorov-Smirnov test to test between two distribution, exact p-value.
static double[]	kolmogorov_smirnov_test(double[] X, double[] Y, TestKind kind, boolean isExact) Compute the Kolmogorov-Smirnov test to test between two distribution.
static double[]	kolmogorov_smirnov_test(double[] X, GenericDistribution dist) Compute the Kolmogorov-Smirnov test to test between X and a known reference distribution, two-sided, exact p-value.
static double[]	kolmogorov_smirnov_test(double[] X, GenericDistribution dist, boolean isExact) Compute the Kolmogorov-Smirnov test to test between X and a known reference distribution, two-sided.
static double[]	kolmogorov_smirnov_test(double[] X, GenericDistribution dist, TestKind kind) Compute the Kolmogorov-Smirnov test to test between X and a known reference distribution, exact p-value.
static double[]	kolmogorov_smirnov_test(double[] X, GenericDistribution dist, TestKind kind, boolean isExact) Compute the Kolmogorov-Smirnov test to test between X and a known reference distribution.
static double[]	kruskal_wallis_test(double[] x, int[] group) Kruskal-Wallis test
static void	main(java.lang.String[] args)
static double[]	mann_whitney_u_test(double[] x, double[] y, double mu, boolean correction, boolean paired, TestKind kind) Mann-Whitney-U test
static double[]	mood_test(double[] x, double[] y) Performs Mood's two-sample test for a difference in scale parameters.
static double[]	mood_test(double[] x, double[] y, TestKind kind) Performs Mood's two-sample test for a difference in scale parameters.
static double[]	poisson_test(int num_events, double time, double rate, TestKind kind) Performs an exact test of a simple null hypothesis about the rate parameter in Poisson distribution
static double[]	poisson_test(int num_events1, int num_events2, double time1, double time2, double r, TestKind kind) Comparison of Poisson rates
static double[]	t_test_paired(double[] x, double[] y, double mu, TestKind kind) Paired t-test
static double[]	t_test(double[] x, double[] y, double mu, boolean pool_var, TestKind kind) Two sample t-test
static double[]	t_test(double[] x, double mu, TestKind kind) One-sample t-test
static double[]	var_test(double[] x, double[] y, double ratio, TestKind kind) Performs an F test to compare the variances of two samples from normal populations.
static double[]	var_test(double[] x, double[] y, TestKind kind) Performs an F test to compare the variances of two samples from normal populations.
static double[]	wilcoxon_test(double[] x, double mu, boolean correction, TestKind kind) One-sample Wilcoxon test.

Methods inherited from class java.lang.Object

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

Constructor Detail

DistributionTest

public DistributionTest()

Method Detail

kolmogorov_smirnov_test

public static final double[] kolmogorov_smirnov_test(double[] X,
                                                     double[] Y)

Compute the Kolmogorov-Smirnov test to test between two distribution, two-sided, exact p-value. If there are ties, then p-values will be inexact!

Parameters:

X - an array with length of nX

Y - an array with length of nY

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

kolmogorov_smirnov_test

public static final double[] kolmogorov_smirnov_test(double[] X,
                                                     double[] Y,
                                                     boolean isExact)

Compute the Kolmogorov-Smirnov test to test between two distribution, two-sided.

Parameters:

X - an array with length of nX

Y - an array with length of nY

isExact - whether the p-value should be computed with the exact method or not (takes a long time). If there are ties, this option is ignored.

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

kolmogorov_smirnov_test

public static final double[] kolmogorov_smirnov_test(double[] X,
                                                     double[] Y,
                                                     TestKind kind)

Compute the Kolmogorov-Smirnov test to test between two distribution, exact p-value. If there are ties, then p-values will be inexact!

Parameters:

X - an array with length of nX

Y - an array with length of nY

kind - the kind of test {LOWER, GREATER, TWO_SIDED}

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

kolmogorov_smirnov_test

public static final double[] kolmogorov_smirnov_test(double[] X,
                                                     double[] Y,
                                                     TestKind kind,
                                                     boolean isExact)

Compute the Kolmogorov-Smirnov test to test between two distribution.

Parameters:

X - an array with length of nX

Y - an array with length of nY

kind - the kind of test {LOWER, GREATER, TWO_SIDED}

isExact - whether the p-value should be computed with the exact method or not (takes a long time). If there are ties, this option is ignored.

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

kolmogorov_smirnov_test

public static final double[] kolmogorov_smirnov_test(double[] X,
                                                     GenericDistribution dist)

Compute the Kolmogorov-Smirnov test to test between X and a known reference distribution, two-sided, exact p-value. If there are ties, then p-values will be inexact!

Parameters:

X - an array with length of nX

dist - reference distribution

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

kolmogorov_smirnov_test

public static final double[] kolmogorov_smirnov_test(double[] X,
                                                     GenericDistribution dist,
                                                     TestKind kind)

Compute the Kolmogorov-Smirnov test to test between X and a known reference distribution, exact p-value. If there are ties, then p-values will be inexact!

Parameters:

X - an array with length of nX

dist - reference distribution

kind - the kind of test {LOWER, GREATER, TWO_SIDED}

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

kolmogorov_smirnov_test

public static final double[] kolmogorov_smirnov_test(double[] X,
                                                     GenericDistribution dist,
                                                     boolean isExact)

Compute the Kolmogorov-Smirnov test to test between X and a known reference distribution, two-sided.

Parameters:

X - an array with length of nX

dist - reference distribution

isExact - whether the p-value should be computed with the exact method or not (takes a long time). If there are ties, this option is ignored.

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

kolmogorov_smirnov_test

public static final double[] kolmogorov_smirnov_test(double[] X,
                                                     GenericDistribution dist,
                                                     TestKind kind,
                                                     boolean isExact)

Compute the Kolmogorov-Smirnov test to test between X and a known reference distribution.

Parameters:

X - an array with length of nX

dist - reference distribution

kind - the kind of test {LOWER, GREATER, TWO_SIDED}

isExact - whether the p-value should be computed with the exact method or not (takes a long time). If there are ties, this option is ignored.

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

ansari_bradley_test

public static final double[] ansari_bradley_test(double[] x,
                                                 double[] y,
                                                 boolean force_exact)

Return the two-sided test of Ansari-Bradley.

Parameters:

x - the original x

y - the original y

force_exact - Set to true if you want exact answer. The default behavior is that if there are ties or either the length of x or the length of y is at least 50.

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

ansari_bradley_test

public static final double[] ansari_bradley_test(double[] x,
                                                 double[] y,
                                                 boolean force_exact,
                                                 TestKind kind)

Ansari-Bradley test.

Parameters:

x - the original x

y - the original y

force_exact - Set to true if you want exact answer. The default behavior is that if there are ties or either the length of x or the length of y is at least 50.

kind - the kind of test {LOWER, GREATER, TWO_SIDED}

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

mood_test

public static final double[] mood_test(double[] x,
                                       double[] y)

Performs Mood's two-sample test for a difference in scale parameters. Two-sided test.

Parameters:

x -

y -

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

mood_test

public static final double[] mood_test(double[] x,
                                       double[] y,
                                       TestKind kind)

Performs Mood's two-sample test for a difference in scale parameters.

Parameters:

x -

y -

kind - the kind of test {LOWER, GREATER, TWO_SIDED}

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

var_test

public static final double[] var_test(double[] x,
                                      double[] y,
                                      TestKind kind)

Performs an F test to compare the variances of two samples from normal populations. Ratio is set to 1.0.

Parameters:

x -

y -

kind - the kind of test {LOWER, GREATER, TWO_SIDED}

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

var_test

public static final double[] var_test(double[] x,
                                      double[] y,
                                      double ratio,
                                      TestKind kind)

Performs an F test to compare the variances of two samples from normal populations.

Parameters:

x -

y -

ratio - the hypothesized ratio of the population variances of x and y.

kind - the kind of test {LOWER, GREATER, TWO_SIDED}

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

wilcoxon_test

public static final double[] wilcoxon_test(double[] x,
                                           double mu,
                                           boolean correction,
                                           TestKind kind)

One-sample Wilcoxon test. Test whether the vector of x is != mu

Parameters:

x -

mu -

correction - set to true if continuity correction is desired. Only matters if x has zeroes or ties

kind - the kind of test {LOWER, GREATER, TWO_SIDED}

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

mann_whitney_u_test

public static final double[] mann_whitney_u_test(double[] x,
                                                 double[] y,
                                                 double mu,
                                                 boolean correction,
                                                 boolean paired,
                                                 TestKind kind)

Mann-Whitney-U test

Parameters:

x -

y -

mu -

correction - set to true if continuity correction is desired. Only matters then there are ties

paired - set to true for paired test (which reduces to Wilcoxon test)

kind - the kind of test {LOWER, GREATER, TWO_SIDED}

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

t_test

public static final double[] t_test(double[] x,
                                    double mu,
                                    TestKind kind)

One-sample t-test

Parameters:

x -

mu -

kind - the kind of test {LOWER, GREATER, TWO_SIDED}

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

t_test_paired

public static final double[] t_test_paired(double[] x,
                                           double[] y,
                                           double mu,
                                           TestKind kind)

Paired t-test

Parameters:

x -

y -

mu -

kind - the kind of test {LOWER, GREATER, TWO_SIDED}

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

t_test

public static final double[] t_test(double[] x,
                                    double[] y,
                                    double mu,
                                    boolean pool_var,
                                    TestKind kind)

Two sample t-test

Parameters:

x -

y -

mu -

pool_var - set to true if the variance should be pooled. Only matters when paired == false

kind - the kind of test {LOWER, GREATER, TWO_SIDED}

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

binomial_test

public static final double[] binomial_test(int n_success,
                                           int n,
                                           double p,
                                           TestKind kind)

Binomial test

Parameters:

n_success - The number of successes

n - The total number of trials

p - Expected probability

kind - the kind of test {LOWER, GREATER, TWO_SIDED}

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

bartlett_test

public static final double[] bartlett_test(double[] x,
                                           int[] group)

Bartlett's test

Parameters:

x -

group - an array of group indices. Observation in x that belongs in the same group must have the same index.

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

fligner_test

public static final double[] fligner_test(double[] x,
                                          int[] group)

Fligner-Killeen test

Parameters:

x -

group - an array of group indices. Observation in x that belongs in the same group must have the same index.

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

kruskal_wallis_test

public static final double[] kruskal_wallis_test(double[] x,
                                                 int[] group)

Kruskal-Wallis test

Parameters:

x -

group - an array of group indices. Observation in x that belongs in the same group must have the same index.

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

poisson_test

public static final double[] poisson_test(int num_events,
                                          double time,
                                          double rate,
                                          TestKind kind)

Performs an exact test of a simple null hypothesis about the rate parameter in Poisson distribution

Parameters:

num_events - number of events.

time - time base for event count.

rate - hypothesized rate

kind - the kind of test {LOWER, GREATER, TWO_SIDED}

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

poisson_test

public static final double[] poisson_test(int num_events1,
                                          int num_events2,
                                          double time1,
                                          double time2,
                                          double r,
                                          TestKind kind)

Comparison of Poisson rates

Parameters:

num_events1 - number of events for the treatment.

num_events2 - number of events for control.

time1 - time base for event count for treatment.

time2 - time base for event count for control.

kind - the kind of test {LOWER, GREATER, TWO_SIDED}

Returns:

an array of two elements: The first is the test statistic, the second is the p-value

diptest

public static final double[] diptest(double[] x)

Perform Hartigan's dip test, assuming the minimum test statistics D is zero.

Parameters:

x - Can be of any order. If x is already sorted, use diptest_presorted to save some time.

Returns:

an array of four elements: The first is the test statistic, the second is the p-value, followed by indices for which there are a dip. If there is no dip, the indices will be set to -1.

diptest_presorted

public static final double[] diptest_presorted(double[] x)

Perform Hartigan's dip test, assuming the minimum test statistics D is zero.

Parameters:

x - MUST BE SORTED in order to output the right result. This routine will NOT check for order!

Returns:

an array of four elements: The first is the test statistic, the second is the p-value, followed by indices for which there are a dip. If there is no dip, the indices will be set to -1.

main

public static final void main(java.lang.String[] args)