Statsmodels

When you need to fit OLS, logistic regression, ARIMA models, or any serious statistical analysis with proper inference tables, this is your tool. It gives you coefficient estimates, p-values, confidence intervals, and all the diagnostic tests you'd expect for publication-quality work. The skill covers linear models, GLMs across different distributions, discrete choice models, and time series. It's heavier than scikit-learn but that's the point: you get rigorous econometric analysis with residual diagnostics, heteroskedasticity tests, and marginal effects. If you're doing statistical modeling where the inference matters as much as the predictions, this is the standard library for it.