RM.ZOIP: ZOIP regression with fixed efects

In ZOIP: ZOIP Distribution, ZOIP Regression, ZOIP Mixed Regression

Description

The RM.ZOIP function adjusts a ZOIP regression model via maximum likelihood. the model may or may not include covariables in any of its parameters, just as it can be a bilaterally inflated model, unilaterally or without parameters of inflation

Usage

RM.ZOIP(formula.mu, formula.sigma = ~1, formula.p0 = ~1, formula.p1 = ~1,
  data, link = c("identity", "identity", "identity", "identity"),
  family = "R-S", optimizer = "nlminb")

Arguments

formula.mu	Formula that defines the regression function for mu, p.e and ~ x1 + x2, it is necessary to define the response variable.
formula.sigma	Formula that defines the regression function for the sigma parameter, a possible value is ~ x1, by default ~ 1.
formula.p0	Formula that defines the regression function for p0, a possible value is ~ x1, by default ~ 1.
formula.p1	Formula that defines the regression function for p1, a possible value is ~ x1, by default ~ 1.
data	It is the data set in data.frame format where it must contain the names of the columns as they are in the formulas.
link	It is a vector with the appropriate link functions for each parameter to be estimated according to the options chosen in the family and formula parameters. If the regression model does not have covariables, the identity option should be used as a link function, regardless of the value chosen in the family, possible options are logit, log, default link = c ('identity', 'identity', 'identity', 'identity').
family	choice of the parameterization or distribution, family = 'R-S' parameterization beta distribution Rigby and Stasinopoulos, 'F-C' distribution Beta parametrization Ferrari and Cribari-Neto, 'Original' Beta distribution classic parameterization, 'Simplex' simplex distribution.
optimizer	Choice of the optimizer, used to find the convergence of the maximum likelihood. you can choose the value of 'nlminb' or 'optim', by default 'nlminb'.

Examples

#Test 1--------------------------------------------------
library(ZOIP)
library(boot)
library(numDeriv)
n<-1000
x1<-stats::runif(n)
x2<-stats::runif(n)

b1<-0.3
b2<--0.5
b3<-0.9
sigma_i<-exp(b1+b2*x1+b3*x2)

c1<-0.2
c2<--1
c3<-0.1
mu_i<-exp(c1+c2*x1)

d1<-0.07
p0_i<-rep(d1,length(n))

e1<-0.02
e2<--4
p1_i<-boot::inv.logit(e1+e2*x2)

param<-cbind(mu_i,sigma_i,p0_i,p1_i)

system.time(y_i<-apply(param,1,function(x){rZOIP(1,mu=x[1],sigma=x[2],
                                                 p0=x[3],p1=x[4],family='Original')}))
data<-as.data.frame(cbind(y_i,x1,x2))

formula.mu=y_i~x1
formula.sigma=~x1+x2
formula.p0=~1
formula.p1=~x2
link=c('log','log','identity','logit')
family='Original'
mod<-RM.ZOIP(formula.mu=formula.mu,formula.sigma=formula.sigma,formula.p0=formula.p0,
             formula.p1=formula.p1,data=data,link=link,family=family)
mod
summary(mod)

ZOIP documentation built on May 1, 2019, 9:11 p.m.