statsmodels.discrete.discrete_model.ProbitResults.resid_dev¶

ProbitResults.resid_dev¶

Deviance residuals

Notes

Deviance residuals are defined

d_{j} = \pm {(2 [Y_{j} \ln (\frac{Y_{j}}{M_{j} p_{j}}) + (M_{j} - Y_{j} \ln (\frac{M_{j} - Y_{j}}{M_{j} (1 - p_{j})})])}^{1 / 2}

where

$p_{j} = c d f (X β)$ and $M_{j}$ is the total number of observations sharing the covariate pattern $j$ .

For now $M_{j}$ is always set to 1.