Fit a BIc (zero-or-one inflated beta) regression

Thin wrapper around gamlss::gamlss() for the inflated beta regression of Ospina and Ferrari (2012), fitting the three (or four, for two-boundary responses) linked submodels of Eq. (1) in Ospina (2026): a discrete inflation-probability submodel and a continuous-mean submodel, optionally with their own precision submodel.

Usage

bic_fit(
  formula,
  sigma.formula = ~1,
  nu.formula = ~1,
  tau.formula = NULL,
  data,
  weights = NULL,
  ...
)

Arguments

formula

Formula for the continuous-mean submodel (mu), e.g. y ~ T + x1 + x2.

sigma.formula

Formula for the precision submodel (default ~1, constant precision).

nu.formula

Formula for the inflation-at-zero submodel.

tau.formula

Formula for the inflation-at-one submodel. Leave NULL for a one-boundary (inflation at zero only) response, fit with gamlss.dist::BEZI; supply it for a two-boundary response, fit with gamlss.dist::BEINF.

data

A data frame.

weights

Optional prior weights (e.g. survey expansion weights), passed to gamlss.

...

Further arguments passed to gamlss::gamlss().

Value

A fitted gamlss object, as returned by gamlss::gamlss(), with its family ("BEZI" or "BEINF") recorded for use by bic_gcomp() and bic_curvature().

References

Ospina, R. and Ferrari, S. L. P. (2012). A general class of zero-or-one inflated beta regression models. Computational Statistics & Data Analysis, 56, 1609-1623.