In MichaelChaoLi-cpu/GWPR.light: Geographically Weighted Panel Regression

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

GWPR.light

This package is grounded in a branch of spatial statistics. Using geographically weights, the geographically weighted panel regression is try to solve the residuals from panel regression clustering spatially. To investigate whether the residuals cluster spatially, the Moran's I test is also improved. Furthermore, three local statistic tests are contained to help the users select model.
This package includes the function for the optimal bandwidth selection in GWPR, the function for GWPR, the function for the local Hausman test, the function for the local F test for individual effects, the function for the local Lagrange Multiplier Breusch-Pagan test, and the function for panel Moran's I test. The functions have been optimized, which require the less memory in the calculation.

Author

Chao Li [e-mail] Shunsuke Managi [e-mail]

Maintainer

Chao Li [e-mail]

Installation

You can install the released version of GWPR.light from CRAN with:

install.packages("GWPR.light")

Detailed Introduction

You can read our vignettes

Example

This is a basic example which shows you how to solve a common problem:

library(GWPR.light)

library(tmap)
## basic example code
data(TransAirPolCalif)
data(California)
formula.GWPR <- pm25 ~ co2_mean + Developed_Open_Space_perc + Developed_Low_Intensity_perc +
   Developed_Medium_Intensity_perc + Developed_High_Intensity_perc +
   Open_Water_perc + Woody_Wetlands_perc + Emergent_Herbaceous_Wetlands_perc +
   Deciduous_Forest_perc + Evergreen_Forest_perc + Mixed_Forest_perc +
   Shrub_perc + Grassland_perc + Pasture_perc + Cultivated_Crops_perc +
   pop_density + summer_tmmx + winter_tmmx + summer_rmax + winter_rmax

This is an example about GWPR.moran.test:

pdata <- plm::pdata.frame(TransAirPolCalif, index = c("GEOID", "year"))
moran.plm.model <- plm::plm(formula = formula.GWPR, data = pdata, model = "within")
#summary(moran.plm.model)

bw.AIC.F <- bw.GWPR(formula = formula.GWPR, data = TransAirPolCalif, index = c("GEOID", "year"), SDF = California,
                     adaptive = F, p = 2, bigdata = F, effect = "individual",
                     model = "within", approach = "AIC", kernel = "bisquare", longlat = F,
                     doParallel = T, cluster.number = 4)

# moran's I test
GWPR.moran.test(moran.plm.model, SDF = California, bw = bw.AIC.F, kernel = "bisquare",
                 adaptive = F, p = 2, longlat=F, alternative = "greater")
# The statistic is significantly greater than 0. Therefore, the residuals are spatially clustered.

GWPR example:

result.F.AIC <- GWPR(bw = bw.AIC.F, formula = formula.GWPR, data = TransAirPolCalif, index = c("GEOID", "year"),
                     SDF = California, adaptive = F, p = 2, effect = "individual", model = "within",
                     kernel = "bisquare", longlat = F)
summary(result.F.AIC$SDF$Local_R2)
tm_shape(result.F.AIC$SDF) +
  tm_polygons(col = "Local_R2", pal = "Reds",auto.palette.mapping = F,
              style = 'cont')

Example of F test:

GWPR.pFtest.resu.F <- GWPR.pFtest(formula = formula.GWPR, data = TransAirPolCalif, index = c("GEOID", "year"),
                                  SDF = California, bw = bw.AIC.F, adaptive = F, p = 2, effect = "individual",
                                  kernel = "bisquare", longlat = F)
tm_shape(GWPR.pFtest.resu.F$SDF) +
     tm_polygons(col = "p.value", breaks = c(0, 0.05, 1))

Example of Locally Breusch-Pagan Lagrange Multiplier Test:

GWPR.plmtest.resu.F <- GWPR.plmtest(formula = formula.GWPR, data = TransAirPolCalif, index = c("GEOID", "year"),
                                    SDF = California, bw = bw.AIC.F, adaptive = F, p = 2,
                                    kernel = "bisquare", longlat = F)
tm_shape(GWPR.plmtest.resu.F$SDF) +
     tm_polygons(col = "p.value", breaks = c(0, 0.05, 1))

Example of Locally Hausman Test Based on GWPR

GWPR.phtest.resu.F <- GWPR.phtest(formula = formula.GWPR, data = TransAirPolCalif, index = c("GEOID", "year"),
                                  SDF = California, bw = bw.AIC.F, adaptive = F, p = 2, effect = "individual",
                                  kernel = "bisquare", longlat = F, random.method = "amemiya")
tm_shape(GWPR.phtest.resu.F$SDF) +
     tm_polygons(col = "p.value", breaks = c(0, 0.05, 1))

MichaelChaoLi-cpu/GWPR.light documentation built on Sept. 23, 2022, 5:25 p.m.