statsmodels.tools.eval_measures.aic¶

statsmodels.tools.eval_measures.aic(llf, nobs, df_modelwc)[ソース]¶

Akaike information criterion

Parameters:¶

llf{float, array_like}

value of the loglikelihood

nobsint

number of observations

df_modelwcint

number of parameters including constant

Returns:¶

aicfloat

information criterion

References

https://en.wikipedia.org/wiki/Akaike_information_criterion