statsmodels

OLS regression with diagnostic tests is executed by fitting linear models using statsmodels, which provides robust inference tools and extensive diagnostics like residual analysis and model selection for econometric applications.

The best way to fit time-series models for forecasting is using statsmodels to estimate ARIMA and VAR frameworks, enabling formal statistical modeling and robust diagnostics for accurate time-dependent predictions.

Generalized linear models handle non-normal outcome variables through statsmodels GLM implementations, supporting continuous and non-normal data distributions for rigorous statistical inference across social science research.

Model selection for discrete-choice models like Logit, Probit, and MNLogit is performed using statsmodels, providing formal modeling frameworks and diagnostic tools to guide inference and reporting for categorical outcomes.

Statsmodels supports robust diagnostics for publication-ready outputs by offering comprehensive model-selection tools and statistical tests that guide formal inference and reporting across linear, GLM, and time-series methods.

WLS or GLS should be used instead of standard OLS when dealing with heteroskedasticity or correlated errors, as statsmodels provides these linear model variants to ensure robust inference and accurate econometric diagnostics.