| mle.zigp {ZIGP} | R Documentation |
'mle.zigp' is used to calculate the MLEs of the regression parameters for mean, overdispersion and zero-inflation.
mle.zigp(Yin, Xin, Win=NULL, Zin=NULL, Offset = rep(1, length(Yin)), init = TRUE)
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'. |
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.
# 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