SpatialRoMLE (version 0.1.0)

RoMLE.error: Robust Maximum Likelihood Estimation for Spatial Error Model

Description

This package provides robust maximum likelihood estimation for spatial error model.

Usage

RoMLE.error(
  initial.beta,
  initial.s2,
  initial.lambda,
  W,
  y,
  x,
  phi.function,
  converge.v,
  iter,
  print.values
)

Arguments

initial.beta

initial value of coefficients

initial.s2

initial value of varaince

initial.lambda

initial value of autocorrelation parameters

W

a symmetric weight matrix

y

dependent variable

x

independent variables

phi.function

a robust m-estimator function, should be set as 1 for Cauchy, 2 for Welsch, 3 for Insha and 4 for Logistic

converge.v

converge value for fisher scoring algorithm, can be set as 1e-04

iter

iteration number for fisher scoring algorithm, set by users (e.g. 100)

print.values

printing estimated values for each step until converge, should be set TRUE or FALSE

Value

coefficients, lambda, s2, Phi

References

Yildirim, V. and Kantar, Y.M. (2020). Robust estimation of spatial error model. in Journal of Statistical Computation and Simulation https://doi.org/10.1080/00949655.2020.1740223

Yildirim, V., Mert Kantar, Y. (2019). Spatial Statistical Analysis of Participants in The Individual Pension System of Turkey. Eskisehir Teknik Universitesi Bilim Ve Teknoloji Dergisi B - Teorik Bilimler, 7(2), 184-194 https://doi.org/10.20290/estubtdb.518706

Examples

# NOT RUN {
#spdep library can be used to create a weight matrix from listw
#require(spdep)
#W <- as(listw, "CsparseMatrix")

#example 1
data(TRQWM)
data(unemployment_data)
data(unemployment_coefs)

y <- unemployment_data$unemployment
x <- unemployment_data$urbanization

#initial values was taken from MLE
initial.beta <- unemployment_coefs[1:2,2]
initial.lambda <- unemployment_coefs[3,2]
initial.s2 <- unemployment_coefs[4,2]

RoMLE.error(initial.beta, initial.s2, initial.lambda, W=TRQWM, y, x,
            phi.function=3, converge.v=0.0001, iter=100, print.values=TRUE)

#example 2
data(TRQWM)
data(IPS_data)
data(IPS_coefs)
y <- IPS_data[,3]
x <- IPS_data[,4:10]

#initial values was taken from MLE
initial.beta <- IPS_coefs[1:8,2]
initial.lambda <- IPS_coefs[9,2]
initial.s2 <- IPS_coefs[10,2]
RoMLE.error(initial.beta, initial.s2, initial.lambda, W=TRQWM, y, x,
            phi.function=3, converge.v=0.0001, iter=100, print.values=TRUE)

# }