StatsModels

Statsmodels is an open-source Python library for statistical modeling and econometrics. It provides classes and functions for estimating a wide range of statistical models, conducting statistical tests, and performing data exploration. The project emphasizes statistical inference, hypothesis testing, and model diagnostics rather than end-to-end prediction alone.

Built on NumPy, SciPy, and Pandas, it supports a formula-like interface via Patsy, allowing users to specify models with readable formulas. It includes linear models (ordinary least squares, weighted least squares, generalized least squares), generalized linear models (logistic, probit, Poisson, gamma, and related families), discrete choice models (Logit, Probit, Multinomial Logit), count models (Poisson, Negative Binomial), and robust regression. It also provides mixed-effects models (MixedLM) and a set of econometric tools for inference and testing.

In time series analysis, statsmodels offers autoregressive and moving-average models, state-space representations, and specialized time-series models such as ARIMA, SARIMAX, and VAR-type models, along with diagnostic tests and information criteria.

Users interact with statsmodels through results objects that expose summaries, confidence intervals, p-values, influence measures, and diagnostic plots. The library integrates with NumPy, SciPy, and Pandas, and leverages the Patsy formula language for model specification. It is BSD-licensed, well-documented, and widely used in academia and industry for statistical inference and methodological research.