There is a nice example of how statsmodels in Python can be used to fit regression equations: https://www.statsmodels.org/stable/regression.html

However, I could not match how the confidence interval for each coefficient estimate is calculated e.g. for coefficient of GPA, the lower and upper bounds should be

0.4639 - 1.96 * 0.162 = 0.14638 and 0.4639   1.96 * 0.162 = 0.78142

But I see 0.132 & 0.796 respectively.

Is that only due to numerical accuracy?

CodePudding user response：

Confidence intervals for coefficient estimates are calculated using Student's t-distribution, not standard normal distribution which you used, hence the mismatch.