In fbahoken/cartogRaflow: Filtering Matrix for Flow Mapping

This vignette of Cartograflow is dedicated to the concentration analysis of an origin-destination matrix for flow mapping purposes.

Main functions

flowgini

-- flowgini() performs a Gini's concentration analysis of the flow features, by computing Gini coefficient and plotting interactive Lorenz curve.

To be use before flowanalysis()

flowanalysis Flow filtering according to a concentration criterion

-- flowanalysis() computes filters criterions based on:

argument critflow is to filter the flows according to their significativity (% of total of flow information) ; \cr
argument critlink is to filter the flows according to their density (% of total features)\cr

These arguments can be used as filter criterion in flowmap().

Useful additional function for a complete analysis

Flowmap

-- flowmap() is to plot flows as segments or arrows, by acting on the following arguments:\cr

filter is to filter or not flow's information or features \cr
threshold is used to set the filtering level of the flows when filter="True" \cr
taille is the value of the width of the flow feature \cr
a.head is the arrow head parameter (in, out, in and out) \cr
a.length is the length of the edges of the arrow head (in inches) \cr
a.angle is the angle from the shaft of the arrow to the edge of the arrow head \cr
a.col is the arrow's color \cr
plota is to add spatial features as map background to the flows's plot \cr
add is to allow to overlay flow features on external spatial features background \cr

Example

rm(list=ls())

library(sf)
library(dplyr)
library(cartograflow)
library(cartography)

1. Load datasets

Flow dataset

# Load Statistical information
data<-read.csv2("./data/MOBPRO_ETP.csv",
                header=TRUE,
                sep=";",
                stringsAsFactors=FALSE,
                encoding="UTF-8",
                dec=".",
                check.names=FALSE)
str(data)

# Variable typing
data$i<-as.character(data$i)
data$j<-as.character(data$j)
data$Fij<-as.numeric(data$Fij)
data$count<-as.numeric(data$count)

# Selecting useful variables
tabflow<-data%>%select(i,j,Fij)

Geographical dataset

# Load a list of geo codes
ID_CODE<-read.csv2("./data/COD_GEO_EPT.csv",
                   header=TRUE,
                   sep=";",
                   stringsAsFactors=FALSE,
                   encoding="UTF-8",
                   dec=".",
                   check.names=FALSE)
#head(ID_CODE)

CODE<-ID_CODE%>% dplyr::select(COD_GEO_EPT)

colnames(CODE)<-c("CODGEO")
#head(CODE)

map<-st_read("./data/MGP_TER.shp")

2. Compute concentration criterion

2.1. Compute the concentration of flow values

# Computes Gini's coefficent
# head(tabflow,3)

tabgini<-flowgini(ODpts = tabflow,
                  origin="i",destination = "j",valflow = "Fij",
                  lorenz.plot = FALSE)

# Interpretation ; The flows are quite concentrated on a few links, the Gini coefficent is equal to 73.16% 

# Plot an interactive Lorenz curve

head(tabgini)

flowgini(ODpts = tabflow,
         origin="i",destination = "j",valflow = "Fij",
         lorenz.plot = TRUE)

2.2. Compute the "critflow" parameter (ex. significance)

# Compute critflow parameter

flowanalysis(tabgini,
             critflow = 0.8,
             result = "signif")

# Interpretation : Flow values up to 13442 are the 80% largest one corresponding to 23,14 % of the total links' features.

2.3. Flowmap filtered according to flows values significance Using the 'critflow" to filter flows

# Graphic parameters
knitr::opts_chunk$set(fig.width=6, fig.height=6)
par(mar=c(0,0,1,0))
extent <- c(2800000, 1340000, 6400000, 4800000)
resolution<-150

# Overlay a spatial background 
par(bg = "NA")
#plot(st_geometry(map), col = "#67786c")

# Add the corresponding background 
plot(st_geometry(map), col = NA, border=NA, bg="#dfe6e1")
plot(st_geometry(map), col = "light grey", add=TRUE)

# For mapping flow up to 11238

flowmap(tab=tabflow,
        fij="Fij",origin.f = "i",destination.f = "j",
        bkg = map,code="EPT_NUM",nodes.X="X",nodes.Y = "Y",
        add=TRUE,
        filter=TRUE,
        threshold=13442,    
        taille=15,           
        a.head = 1,
        a.length = 0.11,
        a.angle = 30,
        a.col="#138913")


# Map Legend
legendPropLines(pos="topleft",
                title.txt="Commuters up to 13442\n (80% of the largest flows)",
                title.cex=0.8,   
                cex=0.5,
                values.cex= 0.7,  
                var=c(13442,max(tabflow$Fij)), 
                lwd=15, 
                frame = FALSE,
                col="#138913",
                values.rnd = 0
                )

#Map cosmetic

layoutLayer(title = "Significant professional mobility in Greater Paris",
           coltitle ="black",
           author = "Cartograflow, 2020",
           sources = "Data : INSEE, 2017 ; Basemap : APUR, RIATE, 2018.",
           scale = 2,
           tabtitle = FALSE,
           frame = TRUE,
           col = "grey",
            )
# north arrow
north("topright")

-- 2.4. Flowmap filtered according to flow features' density Using the 'critlink" to filter flows

flowanalysis(tabgini,
             critlink = 0.1,
             result = "density")

# Interpretation : Flows up to 45772 are the 58.12% largest one corresponding to 10 % of the total links

# Graphic parameters
knitr::opts_chunk$set(fig.width=6, fig.height=6)
par(mar=c(0,0,1,0))
extent <- c(2800000, 1340000, 6400000, 4800000)
resolution<-150

# Overlay a spatial background 
par(bg = "NA")

# Add the corresponding background 
plot(st_geometry(map), col = NA, border=NA, bg="#dfe6e1")
plot(st_geometry(map), col = "light grey", add=TRUE)


# For mapping 10% of the total features as flows up to 45772 commuters

flowmap(tab=tabgini,
        fij="Fij",origin.f = "i",destination.f = "j",
        bkg = map,code="EPT_NUM",nodes.X="X",nodes.Y = "Y",
        add=TRUE,
        plota = TRUE,
        filter=TRUE,
        threshold=45772,    
        taille=15,           
        a.head = 1,
        a.length = 0.15,
        a.angle = 30,
        a.col="#138913")

# Map Legend
legendPropLines(pos="topleft",
                title.txt="Number of commuters up to 45772\n(10% of the links)",
                title.cex=0.8,   
                cex=0.5,
                values.cex= 0.7,  
                var=c(45772,max(tabgini$Fij)), 
                lwd=15, 
                frame = FALSE,
                col="#138913",
                values.rnd = 0
                )

#Map cosmetic
layoutLayer(title = "Low density professional mobility in Greater Paris",
           coltitle ="black",
           author = "Cartograflow, 2020",
           sources = "Data : INSEE, 2017 ; Basemap : APUR, RIATE, 2018.",
           scale = 2,
           tabtitle = FALSE,
           frame = TRUE,
           col = "grey"
            )

# north arrow
north("topright")

Sample datasets

-- Statistical dataset : - INSEE - Base flux de mobilité (2015) -URL : https://www.insee.fr/fr/statistiques/fichier/3566008/rp2015_mobpro_txt.zip

-- Geographical dataset :

municipalities : IGN, GEOFLA 2015 v2.1
Greater Paris : APUR, UMS 2414 RIATE, 2018.

Reference

-- Bahoken Francoise (2016), Programmes pour R/Rtudio annexés, in : Contribution à la cartographie d'une matrix de flux, Thèse de doctorat, Université Paris 7, pp. 325-346. URL : https://halshs.archives-ouvertes.fr/tel-01273776, pp. 480-520.