Statsmodels 均值差异的置信区间

Question

I wanted to find the confidence interval for the difference between two means (Male vs Female). I browsed on the index for statsmodels and found the function below. However it didn't explain where should I specify the Male and Female series. Please advise.

Function:

CompareMeans.tconfint_diff(alpha=0.05, alternative='two-sided', usevar='pooled')

Documentation: https://www.statsmodels.org/stable/generated/statsmodels.stats.weightstats.CompareMeans.tconfint_diff.html

Answer 1

The descriptive statistics of the two series should be passed to the CompareMeans class in DescrStatsW format. After that you can use the tconfint_diff method of the CompareMeans class to obtain the confidence interval for the difference in means.

import pandas as pd
import numpy as np
from statsmodels.stats.weightstats import DescrStatsW, CompareMeans

df = pd.DataFrame({
    'Male': np.random.normal(loc=50, scale=5, size=100),
    'Female': np.random.normal(loc=50, scale=25, size=100),
})

cm = CompareMeans(d1=DescrStatsW(data=df['Male']), d2=DescrStatsW(data=df['Female']))

lower, upper = cm.tconfint_diff(alpha=0.05, alternative='two-sided', usevar='unequal')