statsmodels.discrete.discrete_model.LogitResults.bic

LogitResults.bic

Bayesian information criterion. -2*llf + ln(nobs)*p where p is the number of regressors including the intercept.