DZEXPM (version 2.0)

dzexpm: The estimation and prediction function for the skewed spatial processes

Description

This function provides a way to combine the estimation and prediction procedure for the skewed spatial processes.

Usage

dzexpm(y_ful, x_ful, n_ful, n, u1, u2, theta, p, iter, loopiter)

Arguments

y_ful

the full response variable of a dataset

x_ful

the full covariates of a dataset

n_ful

the full number of the spatial locations in a dataset

n

the number of spatial locations being employed to estimate and evaluate the spatial process

u1

horizontal coordinate of the spatial locations

u2

vertical coordinate of the spatial locations

p

is a useful parameter to capture the "skewness" of the process, which belongs in (0,1)

theta

the range parameter of the covariance kernel in the spatial process

iter

the MCMC iterations to update the parameters

loopiter

the burn in size

Value

A date.frame is returned which includes quantities "DIC", "coverage", "bias.med", "mpe.med" and "SD.med", where "DIC" stands for the Deviance Information Criterion (DIC), see Gelman et al. (2004), and "coverage", "bias.med", "mpe.med" and "SD.med" represent the median of prediction coverage, bias, mean prediction error and standard error over MCMC scheme iterations, respectively, for the predictive performance of the skewed spatial process.

Details

This function is generated according to the spatial model, Double Zero Expectile Normal Process with measurement error, which has good prediction performance. See the references for details.

References

Wang, J., Yang, M. and Majumdar, A. (2018). Comparative Study and Sensitivity Analysis of Skewed Spatial Processes. Computational Statistics, 33,75-98.

Gelman, A., Carlin, J., Stern, H. and Rubin, D. (2004). Bayesian Data Analysis: Second Edition. Texts in Statistical Science. CRC Press. ISBN 1-58488-388-X. LCCN 2003051474. MR 2027492.

Examples

# NOT RUN {
  real<- as.matrix(do.call(cbind, MaxTemp03))
  n_ful<- nrow(real)
  y_ful<- real[,2]
  x_ful<- real[,7]
  x_ful<- log(x_ful)
  x_ful<- scale(x_ful)
  u1<- real[,3]
  u2<- real[,4]
  theta<- 2/150000
  n<- 56
  p<- 0.1
 dzexpm(y_ful, x_ful, n_ful, n, u1, u2, theta, p, 80, 20)
# }