README.md

In LAEQ/gwpr: Functions for calculating and mapping a geographically weighted panel regression (GWPR)

Geographically Weighted Panel Regression (GWPR)

Library aims to offer explicit equations for geographically weighted panel regression (GWPR) and R functions to optimize a bandwidth selection in the panel data case and to produce GWPR results and illustrations.

Requirements

R (>= 4.0)

Installation

devtools::install_github("LAEQ/gwpr")

Examples

Some examples how to use differents functions

library(gwpr)
data(USStates)

# Data preparation: ordering and merging states and their weight
USStates@data$id <- c(1:length(unique(USStates@data[,"state"])))
data <- merge(USStates@data, Produc, by="state", all=T)

dMat <- GWmodel::gw.dist(coordinates(USStates), p=2, longlat=F)
Equation <- log(gsp) ~ log(pcap) + log(pc) + log(emp) + unemp


# Bandwidth selection (CV-score)

# Version avec donnees moyennes (sur le temps)
# adaptive bw
bwAVG.A <- bw.avg(formula=Equation, data=data, SDF=USStates, index=c("id","year"), approach="AICc", kernel="bisquare", adaptive=TRUE, p=2, longlat=FALSE, dMat=dMat)
# fixed bw
bwAVG.F <- bw.avg(formula=Equation, data=data, SDF=USStates, index=c("id","year"), approach="AICc", kernel="bisquare", adaptive=FALSE, p=2, longlat=F, dMat=dMat)

# Version avec pseudo-CV
# adaptive bw
bwCV.A <-  bw.CV.A(formula=Equation, data=data, index=c("id","year"), effect='individual', model="within", kernel="bisquare", dMat=dMat, bws=c(30:40))
# fixed bw
bwCV.F <-  bw.CV.F(formula=Equation, data=data, index=c("id","year"), effect='individual', model="within", kernel="bisquare", dMat=dMat, interval=c(1500000, 2500000))

# GWPR model

# L'application empirique avec laquelle nous comparons (panel classique) est celle de
# Baltagi, B. H. (2005) Econometric Analysis of Panel Data (third ed.) qui reprend l'exemple de Munnell (1990)

# individual FE
# adaptive bw
USgwplm.A <- gwplm(SpDF=USStates, data=data, index=c("id", "year"), formula=Equation, bw=bwCV.A, kernel="bisquare", adaptive=TRUE, effect="individual", model="within", dMat=dMat)
# fixed bw
USgwplm.F <- gwplm(SpDF=USStates, data=data, index=c("id", "year"), formula=Equation, bw=bwCV.F, kernel="bisquare", adaptive=FALSE, effect="individual", model="within", dMat=dMat)

Authors

• Maxime Gaboriault - Creator
• Philippe Apparicio - Author
• David Maignan - Compiler

LAEQ/gwpr documentation built on June 28, 2020, 8:23 p.m.