mqr.inference.correlation.

test#

mqr.inference.correlation.test(x, y, H0_corr=0.0, alternative='two-sided')#

Hypothesis test on the Pearson correlation coefficient.

Null-hypothesis

corr(x, y) == H0_corr

Parameters:
x, yarray_like

Test correlation of these two equal-length samples.

H0_corrfloat, optional

Null-hypothesis correlation coefficient.

alternative{‘two-sided’, ‘less’, ‘greater’}, optional

Sense of alternative hypothesis.

Returns:
mqr.inference.hyptest.HypothesisTest
Raises:
ValueError

Length of x and y are different.

Notes

Chooses a method depending on the null-hypothesis. When the null-hypothesis correlation is zero, calculates a t-statistic according to Pearson. On the other hand, when the null-hypothesis correlation is non-zero, calculates a Gaussian statistic based on Fisher’s z-transform (see [1]).

References

[1]

Asuero, A. G., Sayago, A., & González, A. G. (2006). The correlation coefficient: An overview. Critical reviews in analytical chemistry, 36(1), 41-59.