Generalised gamma delay distribution family

A brms custom family for the generalised gamma distribution in the Prentice (1974) parameterisation of flexsurv::dgengamma(). mu and sigma are the location and scale of the log delay and Q is a positive shape. The Weibull (Q = 1) and gamma (Q = sigma) families are special cases and the lognormal is the limit as Q goes to zero. It needs the flexsurv package and works with every epidist model.

Usage

gengamma(link = "identity", link_sigma = "log", link_Q = "log")

Arguments

link, link_sigma, link_Q

The link functions of mu, sigma and Q.

Value

A brms custom family object.

References

Prentice, R. L. (1974). A log gamma model and its maximum likelihood estimation. Biometrika, 61(3), 539-544. doi:10.1093/biomet/61.3.539

See also

Other family: epidist_family(), epidist_family_model(), epidist_family_model.default(), epidist_family_param(), epidist_family_param.default(), epidist_family_param.gengamma()

Examples

gengamma()
#> 
#> Custom family: gengamma 
#> Link function: identity 
#> Parameters: mu, sigma, Q 
#>