statsmodels.stats.weightstats.CompareMeans

class statsmodels.stats.weightstats.CompareMeans(d1, d2)

class for two sample comparison

The tests and the confidence interval work for multi-endpoint comparison: If d1 and d2 have the same number of rows, then each column of the data in d1 is compared with the corresponding column in d2.

Parameters

d1, d2instances of DescrStatsW

Notes

The result for the statistical tests and the confidence interval are independent of the user specified ddof.

TODO: Extend to any number of groups or write a version that works in that case, like in SAS and SPSS.

Methods

dof_satt()	degrees of freedom of Satterthwaite for unequal variance
from_data(data1, data2[, weights1, …])	construct a CompareMeans object from data
summary([use_t, alpha, usevar, value])	summarize the results of the hypothesis test
tconfint_diff([alpha, alternative, usevar])	confidence interval for the difference in means
ttest_ind([alternative, usevar, value])	ttest for the null hypothesis of identical means
ttost_ind(low, upp[, usevar])	test of equivalence for two independent samples, base on t-test
zconfint_diff([alpha, alternative, usevar])	confidence interval for the difference in means
ztest_ind([alternative, usevar, value])	z-test for the null hypothesis of identical means
ztost_ind(low, upp[, usevar])	test of equivalence for two independent samples, based on z-test

Properties

std_meandiff_pooledvar	variance assuming equal variance in both data sets
std_meandiff_separatevar