Reproduction-Material-Spatial-Monte-Carlo-Experiments

This repository provides all materials for reproducing the results of “Spatial Regression Models: A Systematic Comparison of Different Model Specifications using Monte Carlo Experiments”.

Requirements

The codes needs the following folders: “01_Script”, “02_Data”, “03_Output”. All R Scripts are required in folder “01_Script”.

The following packages are necessary for reproduction of main results:

install.packages("ggmap")
install.packages("RgoogleMaps") 
install.packages("foreign")
install.packages("sp") 
install.packages("GISTools")
install.packages("sp") 
install.packages("spdep")
install.packages("texreg") 
install.packages("xtable")
install.packages("rgdal") 
install.packages("raster")
install.packages("extrafont")

Explanations

All R scripts named “01~” contain the functions required for the simulations. To reproduce the results of the final paper, you need to run the R script “02_Monte Carlo Simulation Spatial_Main Analyses”. Results of the Supplement are produced by running “03_Monte Carlo Simulation Spatial_Supplementary Analyses”. You only need to change the superordinate directory “dir” (line 12), containing the three sub-folders. Generated data will be saved in “02_Data”, output will be generated in the “03_Output” directory.

Note that reproducing the results with R=1000 simulation runs may take a lot of time on a conventional computer. Thus, all simulation programs are also available as “~_hpc” version for usage on a high performance cluster with the package “doParallel”. Depending on your system, you need to set up additional batch files to pass the code to a high performance cluster. However, it may decrease simulation time dramatically.

System and version information

Platform: Windows 7 (x86_64-w64-mingw32)

Version: R version 3.5.0