mle.zigp {ZIGP}R Documentation

Maximum Likelihood Estimates

Description

'mle.zigp' is used to calculate the MLEs of the regression parameters for mean, overdispersion and zero-inflation.

Usage

mle.zigp(Yin, Xin, Win=NULL, Zin=NULL, Offset = rep(1, length(Yin)),
init = TRUE)

Arguments

Yin response vector of length n.
Xin design matrix of dim (n x p) for mean modelling.
Win design matrix of dim (n x r) for overdispersion modelling.
Zin design matrix of dim (n x q) for zero inflation modelling.
Offset exposure for individual observation lengths. Defaults to a vector of 1. The offset MUST NOT be in 'log' scale.
init a logical value indicating whether initial optimization values for dispersion are set to -2.5 and values for zero inflation regression parameters are set to -1 (init = F) or are estimated by a ZIGP(mu(i), phi, omega)-model (init = T). Defaults to 'T'.

References

Czado, C., Erhardt, V., Min, A., Wagner, S. (2007) Zero-inflated generalized Poisson models with regression effects on the mean, dispersion and zero-inflation level applied to patent outsourcing rates. Statistical Modelling 7 (2), 125-153.

Examples

# Number of damages in car insurance.
# (not a good fit, just to illustrate how the software is used)
     
damage <- c(0,1,0,0,0,4,2,0,1,0,1,1,0,2,0,0,1,0,0,1,0,0,0)
Intercept <- rep(1,length(damage))
insurance.year <- c(1,1.2,0.8,1,2,1,1.1,1,1,1.1,1.2,1.3,0.9,1.4,1,1,1,1.2,
1,1,1,1,1)
drivers.age <- c(25,19,30,48,30,18,19,29,24,54,56,20,38,18,23,58,
47,36,25,28,38,39,42)
# for overdispersion: car brand dummy in {1,2,3}, brand = 1 is reference
brand <- c(1,2,1,3,3,2,2,1,1,3,2,2,1,3,1,3,2,2,1,1,3,3,2)
brand2 <- ifelse(brand==2,1,0)
brand3 <- ifelse(brand==3,1,0)
# abroad: driver has been abroad for longer time (=1)
abroad <- c(0,0,0,1,0,0,1,0,0,0,0,0,1,0,0,0,0,1,0,1,1,1,1)
Y <- damage
X <- cbind(Intercept, drivers.age)
W <- cbind(brand2,brand3)
Z <- cbind(abroad) # so name will be printed

mle.zigp(Yin=Y, Xin=X, Win=W, Zin=Z, Offset = insurance.year, init = FALSE)

# Output can be summarized as:

#[1]   Range for ZI-Parameters:   0.2491062 0.5
#[2]   Range of Dispersion Pars:  1.000176 2.189325
#[3]   Coefficients for mu:       1.471478 -0.05075418
#[4]   Coefficients for phi:      -8.646371 0.1733860
#[5]   Coefficients for omega:    -1.103385
#[6]   Pearson Chi Squared:       15.09779
#[7]   Range of mu:               0.2294054 2.445806
#[8]   Message:                   "NULL"
#[9]   AIC:                       56.88305

[Package ZIGP version 2.7 Index]