I am comparing stats.ttest_ind() vs "manual" computation of the same test, and get different results.
import numpy as np import pandas as pd import scip