statsmodels.genmod.families.family.InverseGaussian

class statsmodels.genmod.families.family.InverseGaussian(link=<class 'statsmodels.genmod.families.links.inverse_squared'>)

InverseGaussian exponential family.

Parameters:

link : a link instance, optional

The default link for the inverse Gaussian family is the inverse squared link. Available links are inverse_squared, inverse, log, and identity. See statsmodels.family.links for more information.

See also

statsmodels.genmod.families.family.Family, Link Functions

Notes

The inverse Guassian distribution is sometimes referred to in the literature as the Wald distribution.

Attributes

InverseGaussian.link	(a link instance) The link function of the inverse Gaussian instance
InverseGaussian.variance	(varfunc instance) variance is an instance of statsmodels.family.varfuncs.mu_cubed

Methods

deviance(endog, mu[, freq_weights, scale])	Inverse Gaussian deviance function
fitted(lin_pred)	Fitted values based on linear predictors lin_pred.
loglike(endog, mu[, freq_weights, scale])	The log-likelihood function in terms of the fitted mean response.
predict(mu)	Linear predictors based on given mu values.
resid_anscombe(endog, mu)	The Anscombe residuals for the inverse Gaussian distribution
resid_dev(endog, mu[, scale])	Returns the deviance residuals for the inverse Gaussian family.
starting_mu(y)	Starting value for mu in the IRLS algorithm.
variance
weights(mu)	Weights for IRLS steps

Methods

deviance(endog, mu[, freq_weights, scale])	Inverse Gaussian deviance function
fitted(lin_pred)	Fitted values based on linear predictors lin_pred.
loglike(endog, mu[, freq_weights, scale])	The log-likelihood function in terms of the fitted mean response.
predict(mu)	Linear predictors based on given mu values.
resid_anscombe(endog, mu)	The Anscombe residuals for the inverse Gaussian distribution
resid_dev(endog, mu[, scale])	Returns the deviance residuals for the inverse Gaussian family.
starting_mu(y)	Starting value for mu in the IRLS algorithm.
weights(mu)	Weights for IRLS steps