pyhctsa.operations.hypothesis_tests.independence_tests¶

pyhctsa.operations.hypothesis_tests.independence_tests(y, the_test='runsz')¶

Statistic or p-value of a test of serial independence.

Tests whether the values of the time series are independent of one another. Unlike the tests in marginal_tests(), the result depends on the temporal order of the values. This is the part of hctsa’s former HT_HypothesisTest that tests dependence. The runs test is returned as its z-statistic (in closed form), the other tests as p-values.

Parameters:¶

y : array-like¶

The input time series.

the_test : {'runsz', 'runstest', 'lbq'}, optional¶

The test:

• ’runsz’: runs test for randomness of the runs of values above and below the median; returns the signed z-statistic of the number of runs (pyhctsa.robust.bf_runs_z()): negative for fewer runs than expected (positive serial dependence), positive for more (alternation), and approximately standard normal under the null hypothesis;

• ’runstest’: the p-value of the same hypothesis (runs above and below the mean, exact distribution of the number of runs);

• ’lbq’: Ljung-Box Q-test for autocorrelation up to lag 20.

Default is 'runsz'.

Returns:¶

The z-statistic of the test for ‘runsz’, otherwise its p-value: the probability, under the null hypothesis, of a test statistic at least as extreme as that observed. Small values are evidence of serial dependence.

Return type:¶

float