Notes on Chapter 2 - Maps & Graphs
J. Kavanagh
20/2/2018
The first thing to do is load the following packages in R or RStudio.
library(tmap)
library(sp)
library(tmaptools)
library(ggplot2)
library(ggthemes)
library(dplyr)##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionlibrary(seriation)Load the RData file C2.RData
You can find this file using the following link. Download it and place it within your workspace folder for R or RStudio. For instructions on how to set your workspace directory please use the following instructions.
load('C2.RData')This will add all the datasets that were created for this chapter to your R workspace. As there are two samples in this chapter the naming conventions followed this approach: the datasets that are for the Enlisted sample start with ‘enlist’ and the datasets that are for the Officers sample start with ‘off’. For example the date of attesation file for the enlisted sample is called ‘enlist_attest’.
head(enlist_attest)## Date.of.Attestation month
## 1 1922-07-14 7
## 2 1922-07-17 7
## 3 1922-07-19 7
## 4 1922-07-19 7
## 5 1922-07-22 7
## 6 1922-08-01 8Most of the datasets were originally created in wide format files as shown in the example of enlist_rank_comp which refers to the comparison between ranks held in different militaries.
head(enlist_rank_comp)## X Volunteer Private Corporal Sergeant Other Blank
## 1 National Army 46 0 0 3 0 1
## 2 British Army 0 25 9 5 2 1
## 3 Royal Air Force 0 0 0 1 0 1
## 4 Canadian Army 0 0 0 0 0 1
## 5 Not Listed 0 0 0 0 5 0It was necessary to change the file into long form data. This was done using the melt command in the package ‘reshape2’. The result is the dataset enlist_rank_c
require(reshape2)## Loading required package: reshape2enlist_rank_C <- melt(enlist_rank_comp)## Using X as id variableshead(enlist_rank_C)## X variable value
## 1 National Army Volunteer 46
## 2 British Army Volunteer 0
## 3 Royal Air Force Volunteer 0
## 4 Canadian Army Volunteer 0
## 5 Not Listed Volunteer 0
## 6 National Army Private 0This process was repeated for the next enlisted dataset enlist_prevRB, which is a comparison of previous ranks by branch of military service.
head(enlist_prevRB)## X Private Corporal Sergeant Other Blank
## 1 British Army 69 35 16 7 1
## 2 Royal Air Force 0 0 1 0 1
## 3 Royal Marines 0 0 2 0 0
## 4 Royal Navy 0 2 0 0 0
## 5 Canadian Army 0 0 0 0 1
## 6 Not Listed 0 0 0 11 0enlist_prevRank <- melt(enlist_prevRB)## Using X as id variableshead(enlist_prevRank)## X variable value
## 1 British Army Private 69
## 2 Royal Air Force Private 0
## 3 Royal Marines Private 0
## 4 Royal Navy Private 0
## 5 Canadian Army Private 0
## 6 Not Listed Private 0Here is the code for each of the graphs that utilise the three enlisted sample datasets outlined above
Note - Each graph utilises the colorblind setting for maximum accessibility
This is the first graph, showing monthly totals of recruitment.
ggplot(enlist_attest, aes(x=month)) + geom_bar() + scale_x_continuous(breaks = c(7,8,9,10,11,12), labels = c("July", "August", "September", "October", "November", "December")) + labs(title= "Fig. 1 - Date of Attestation by Month Enlisted Sample", caption="Source: Sample of ex-Britishers - Miltiary Archives") + xlab("Monthly Totals") + ylab("Number of Recruits") + theme_economist() + theme(axis.text.x = element_text(colour = "darkslategrey", size = 10), axis.text.y = element_text(colour = "darkslategrey", size = 10), text = element_text(family = "Georgia"),
plot.title = element_text(size = 10, margin = margin(b = 10)),
plot.subtitle = element_text(size = 12, color = "darkslategrey", margin = margin(b = 10)),
plot.caption = element_text(size = 8, margin = margin(t = 10), color = "grey70", hjust = 0))
This is the second plot showing the previous ranks held in the different non-Irish militaries.
bertinplot(enlist_PREVB, options = list(shading=TRUE, panel = panel.bars, reverse = TRUE))
bertinplot(enlist_PREVB, s, options = list(shading=TRUE, panel = panel.bars, reverse = TRUE))
This is the third graph showing the comparison between ranks held in non-Irish militaries and the Irish National Army in 1922.
bertinplot(enlist_RC, options = list(shading=TRUE, panel = panel.bars, reverse = TRUE))
bertinplot(enlist_RC, t, options = list(shading=TRUE, panel = panel.bars, reverse = TRUE))
Repeat the same process used above when analysing the officer sample datasets
Each original dataset has been changed to a long format using the melt() command in the package reshape2.
head(off_rankCC)## X Colonel Commandant Captain Lieutenant X1LT X2LT
## 1 Army Corps of Engineers 0 8 20 10 4 6
## 2 A.G.’s Department 0 1 4 4 1 1
## 3 Army Air Service 0 0 6 2 4 2
## 4 Athlone Command 0 1 10 0 2 2
## 5 Claremorris Command 1 3 9 1 7 4
## 6 Curragh Command 2 1 6 1 3 6
## Cadet NoRank
## 1 1 2
## 2 0 0
## 3 0 0
## 4 0 0
## 5 0 0
## 6 0 0head(off_yrservice)## year variable value
## 1 <1 year IRA 12
## 2 1-5 years IRA 96
## 3 5+ years IRA 2
## 4 10+years IRA 0
## 5 <1 year BA 2
## 6 1-5 years BA 139head(off_recruit_typeX)## X variable value
## 1 ex-British February 0
## 2 IRA & British Service February 11
## 3 ex-British March 4
## 4 IRA & British Service March 16
## 5 ex-British April 4
## 6 IRA & British Service April 13head(off_rankcomp1)## X rank total
## 1 National Army 1922 Colonel 2
## 2 National Army 1923 Colonel 3
## 3 National Army 1922 Commandant 5
## 4 National Army 1923 Commandant 6
## 5 National Army 1922 Brigadier 1
## 6 National Army 1923 Brigadier 0Here is the code used to create the graphs for each of the Officer sample datasets outlined above
This is a bertinplot that shows the amount of officers per command by type. It provides an effective overview of the distribution of the officers sample by specific commands and corps of the National Army. The first plot is ordered by the original file and the second plot is reorded using BEA.
bertinplot(off_rank_C, options = list(shading=TRUE, panel = panel.blocks, reverse = TRUE))
bertinplot(off_rank_C, o, options = list(shading=TRUE, panel = panel.blocks, reverse = TRUE))
This plot shows the amount of previous military service that each officer had listed before joining the National Army.
This graph is an overview of recruitment totals per month for the officers sample for 1922 and 1923.
ggplot(off_recruitX, aes(x=variable, y=value, fill= factor(X))) + geom_bar(stat = "identity", color="grey24", position = "dodge") + labs(fill="Year", labels=c("1922", "1923")) + scale_fill_colorblind() + xlab("Months") + ylab("No of Recruits") + ggtitle("Fig. 6 Dates of Attestation 1922 - 1923") + theme_economist() +
theme(axis.text.x = element_text(colour = "darkslategrey", size = 6), axis.text.y = element_text(colour = "darkslategrey", size = 12), text = element_text(family = "Georgia"),
plot.title = element_text(size = 12, margin = margin(b = 10)),
plot.subtitle = element_text(size = 12, color = "darkslategrey", margin = margin(b = 25)),
plot.caption = element_text(size = 8, margin = margin(t = 10), color = "grey70", hjust = 0))
This graph shows the number of recruits, either ex-British military or ex-IRA & ex-British military, who joined the National Army per month for 1922.
ggplot(off_recruit_typeX, aes(x=variable, y=value, fill= factor(X))) + geom_bar(stat = "identity", color="grey24", position = "dodge") + labs(fill="Year", labels=c("ex-British", "ex-British/IRA")) + scale_fill_colorblind() + xlab("Months") + ylab("No of Recruits") + ggtitle("Fig. 7 Types of Recruits Feb - Dec 1922") + theme_economist() +
theme(axis.text.x = element_text(colour = "darkslategrey", size = 6), axis.text.y = element_text(colour = "darkslategrey", size = 12), text = element_text(family = "Georgia"),
plot.title = element_text(size = 12, margin = margin(b = 10)),
plot.subtitle = element_text(size = 12, color = "darkslategrey", margin = margin(b = 25)),
plot.caption = element_text(size = 8, margin = margin(t = 10), color = "grey70", hjust = 0))
This is a bar chart showing the differences in the types of rank held by a sub-group of the Officers sample who were found listed on the Irish Army Census between 1922 and 1923.
ggplot(off_rankcomp1, aes(x=reorder(rank), y= total, fill= factor(X))) + geom_bar(stat = "identity", position = "dodge") + scale_fill_colorblind(name="Year") + ggtitle("Fig. 8 Comparison of ranks held in 1922 versus 1923 by sample of ex-British Officers") + xlab("Year") + ylab("No. of Officers") + theme_economist() + theme(axis.text.x = element_text(colour = "darkslategrey", size = 4), axis.text.y = element_text(colour = "darkslategrey", size = 8), text = element_text(family = "Georgia"),
plot.title = element_text(size = 12, margin = margin(b = 10)),
plot.subtitle = element_text(size = 12, color = "darkslategrey", margin = margin(b = 25)),
plot.caption = element_text(size = 8, margin = margin(t = 10), color = "grey70", hjust = 0))
Mapping the Samples
Before the samples can be mapped, the totals need to be added to the ESRI shapefile. These totals are contained in a dataset called ‘address_total’
head(address_total)## county ADDR_MIX ADDR_TOTALMIL NOK_TOTALMIL ADDR_ENLIST ADDR_OFF
## 1 County Antrim 5 885 854 104 3
## 2 County Armagh 1 130 125 NA 2
## 3 County Carlow NA 550 542 NA NA
## 4 County Cavan 1 616 596 NA 3
## 5 County Clare 7 1088 1023 NA NA
## 6 County Cork 16 2746 2629 NA 4
## NOK_OFF ADDR_OFFMIL NOK_OFFMIL
## 1 3 17 15
## 2 NA 3 4
## 3 3 19 17
## 4 3 49 45
## 5 NA 77 69
## 6 4 185 173This file is a series of totals collated from the Irish Army Census 1922 and the two samples of ex-British military personnel collated for this chapter. The Irish Army Census 1922 lists three types of addresses: - 1. Home Address - 2. Next of Kin Address - 3. Place of Attestation.
For the purposes of this chapter the Home Adddress and the Next of Kin Address have been summarised into a series of totals by county.
The abbreviations are as follows:
- ADDR_TOTALMIL = Total recruits by County (Mil Census)
- NOK_TOTALMIL = Total recruits by Next of Kin County (Mil Census)
- ADDR_ENLIST = Enlisted sample by county
- ADDR_OFF = Officers sample by county
- NOK_OFF = Officers sample by Next of Kin county
- ADDR_MIX = Mixed sample by county
- ADDR_OFFMIL = Total officers by county (Mil Census)
- saveNOK_OFFMIL = Total officers by Next of Kin county (Mil Census)
A all-Ireland map was created as part of the Townloads project
You can download this file from the following link. The file was reduced in size using QGIS. It is available as part of the C2.RData file under the name ire_counties
This map can be viewed using the R package ‘Tmap’.
Use the qtm command to view the map
qtm(ire_counties)
The next step is to merge the data from ‘addresses_total’ to the map file.
recruit_map <- merge(ire_counties, address_total, by.x="NAME_EN", by.y="county")This creates a new map file called recruit_map and it now has added all the totals to the each of the thirty-two counties in Ireland in 1922.
Next, check that the merge into the new map file, ‘recruit_map’ was successful by running the following code.
head(recruit_map@data)## NAME_EN NAME_TAG ALT_NAME LATITUDE LONGITUDE ADDR_MIX
## 11 County Galway Galway <NA> 53.36274 -8.953548 5
## 26 County Sligo Sligo <NA> 54.15746 -8.623521 1
## 16 County Leitrim Leitrim <NA> 54.13998 -8.022640 1
## 22 County Meath Meath <NA> 53.63153 -6.731971 2
## 9 County Dublin Dublin <NA> 53.39225 -6.283930 38
## 5 County Clare Clare <NA> 52.84052 -9.042776 7
## ADDR_TOTALMIL NOK_TOTALMIL ADDR_ENLIST ADDR_OFF NOK_OFF ADDR_OFFMIL
## 11 1328 1280 NA 2 2 66
## 26 340 337 NA 1 1 38
## 16 323 316 NA NA NA 24
## 22 675 657 NA 2 2 59
## 9 6988 6530 NA 22 22 426
## 5 1088 1023 NA NA NA 77
## NOK_OFFMIL
## 11 65
## 26 36
## 16 24
## 22 58
## 9 416
## 5 69Creating a subset for Ulster Province
As one of the samples of ex-servicemen utilised in this chapter is from Ulster, it was necessary to create a smaller map of Ulster from the larger recruit_map. This was created using the subset.
ulster <- subset(recruit_map, NAME_TAG %in% c("Antrim", "Armagh", "Down", "Donegal", "Fermanagh", "Cavan", "Monaghan", "Londonderry", "Tyrone"))Use the qtm command to check that the map contains all the necessary counties for the province of Ulster
ulster %>% qtm
Inside the new map file of Ulster you will find that all the data from the original map is also present for this province.
ulster@data## NAME_EN NAME_TAG ALT_NAME LATITUDE LONGITUDE ADDR_MIX
## 23 County Monaghan Monaghan <NA> 54.15519 -6.923466 NA
## 10 County Fermanagh Fermanagh <NA> 54.35014 -7.641221 NA
## 7 County Donegal Donegal <NA> 54.92274 -7.905062 3
## 28 County Tyrone Tyrone <NA> 54.61135 -7.160032 1
## 2 County Armagh Armagh <NA> 54.29845 -6.551526 1
## 18 County Londonderry Londonderry County Derry 54.92932 -6.846983 2
## 8 County Down Down <NA> 54.36239 -5.940349 NA
## 4 County Cavan Cavan <NA> 53.99487 -7.352358 1
## 1 County Antrim Antrim <NA> 54.87024 -6.202141 5
## ADDR_TOTALMIL NOK_TOTALMIL ADDR_ENLIST ADDR_OFF NOK_OFF ADDR_OFFMIL
## 23 480 456 NA 1 1 35
## 10 279 270 NA 1 1 13
## 7 1024 1006 NA 2 2 58
## 28 351 340 22 NA NA 9
## 2 130 125 NA 2 NA 3
## 18 262 255 11 NA NA 9
## 8 153 151 7 NA NA 3
## 4 616 596 NA 3 3 49
## 1 885 854 104 3 3 17
## NOK_OFFMIL
## 23 33
## 10 16
## 7 50
## 28 8
## 2 4
## 18 9
## 8 3
## 4 45
## 1 15Here is an all-Ireland map of recruitment from January - November 1922
This was taken from the Irish Army Census totals using the Home Addresses given by each soldier.
tm_shape(recruit_map) + tm_fill(col = "ADDR_TOTALMIL", title = 'National Army Recruitment 1922', style = "cont", n=5) + tm_borders(col = "grey25", lty = "solid", lwd = 2) + tm_style_col_blind() + tm_text("NAME_TAG", size = "AREA", scale = 0.5)
As each soldier gave an individual address, it was possible to tabluate the county totals using a few lines of code. Many thanks to Martin Charlton of the National Centre for Geocomputation (NCG) at Maynooth University for creating this code sequence for processing the Irish Army Census data.
The code is as follows
Counties <- c("Antrim","Armagh", "Carlow", "Cavan", "Clare", "Cork", "Derry", "Donegal", "Down", "Dublin",
"Fermanagh", "Galway", "Kerry", "Kildare", "Kilkenny", "Laois", "Leitrim",
"Limerick", "Longford", "Louth", "Mayo", "Meath", "Monaghan", "Offaly",
"Roscommon", "Sligo", "Tipperary", "Tyrone", "Waterford", "Westmeath", "Wexford",
"Wicklow")
x <- read.csv("cavan.csv", stringsAsFactors=FALSE)
head(x)
Addresses <- x$Home.Address
x$County <- ""
N <- length(Addresses)
for(i in 1:N) {
if(Addresses[i] == "") {
x$County[i] <- "Unknown"
} else {
add.split <- strsplit(Addresses[i], ",")
n <- length(add.split[[1]])
last.bit <- add.split[[1]][n]
last.bit <- gsub(" ","",last.bit)
cty <- match(last.bit,Counties)
if(is.na(cty)) {
x$County[i] <- "Unknown"
} else {
x$County[i] <- Counties[cty]
}
}
}Now there is no reason to run this code section as the totals have already been created, but this was the process that created specific county totals for the twenty-six counties covered by the Irish Army Census. As the Irish Army Census was collated by the Military Archives into county segments, each was exported from the database as a csv file and was processed using the above code for the Home Address and Next of Kin Address field. The ‘unknown’ column which this code creates was then checked by hand in Apple Numbers and manual changes were applied where the county could be ascertained. i.e. Home addresses listing towns such as ‘Ennis’ were re-categorised as County Clare for the purpose of creating these totals.
Ulster Maps
This map is a distribution of the Enlisted sample of ex-servicemen in Ulster.
tm_shape(ulster) + tm_fill(col = "ADDR_ENLIST", colorNA = "grey", title = 'Enlisted Sample Recruitment', style = "cont", n=4) + tm_borders(col = "grey40", lwd = 1) + tm_style_col_blind() + tm_text("NAME_TAG", size = "AREA", scale = 0.5)
This map is a comparative map of overall recruitment in Ulster.
tm_shape(ulster) + tm_fill(col = "ADDR_TOTALMIL", colorNA = "grey", title = 'Total Ulster Province Recruitment', style = "cont", n=4) + tm_borders(col = "grey40", lwd = 1) + tm_style_col_blind() + tm_text("NAME_TAG", size = "AREA", scale = 0.5) 
Here is a comparative map combining the two earlier maps of Ulster recruitment.
tm_shape(ulster) + tm_fill(col=c("ADDR_ENLIST", "ADDR_TOTALMIL"), title = c("Enlisted Sample", "Total Ulster Recruitment"), colorNA = "grey", textNA = "No Enlistments", style = "cont") + tm_borders() + tm_style_col_blind() + tm_scale_bar() + tm_layout(panel.labels = c("Enlisted Sample - Ulster Distribution", "Total Ulster Recruitment - Irish Army Census"))
Here is a comparative map of Home Address and Next of Kin Address as listed on the Irish Army Census. It is clear that there is little to no variation between the two datasets. This shows that there was little variation between the subject and their NOK details annotated on the census.
tm_shape(recruit_map) + tm_fill(col=c("ADDR_TOTALMIL", "NOK_TOTALMIL"), title=c('Map of Home Addresses', 'Map of NOK Home Addresses'), style="cont") + tm_borders() + tm_style_col_blind() + tm_scale_bar()
This map shows the distribution of the Officers sample across the entirety of Ireland. It is quite limited due to the small number of officers that were found on the Irish Army Census.
tm_shape(recruit_map) + tm_fill(col="ADDR_OFF", title='Map of Officers Sample', style="cont", colorNA = "grey", textNA = "Not Listed") + tm_borders() + tm_style_col_blind() + tm_scale_bar()
This map is a comparative map of the Officers sample versus the distribution of the total officer corps across Ireland. The officer corps in this sample is defined as any officer with a rank higher than Sergeant.
tm_shape(recruit_map) + tm_fill(col=c("ADDR_OFF", "ADDR_OFFMIL"), title=c('Map of Officers Sample', 'Map of Mil Census Officer Corps'), style="cont", colorNA = "grey", textNA = "Not Listed") + tm_borders() + tm_style_col_blind() + tm_scale_bar()
This final map is a comparative map showing the distribution of the mixed sample versus the countrywide National Army enlistment
tm_shape(recruit_map) + tm_fill(col=c("ADDR_MIX", "ADDR_TOTALMIL"), title=c('Map of Mixed Sample', 'Map of total National Army Enlistment 1922'), style="cont", colorNA = "grey", textNA = "Not Listed") + tm_borders() + tm_style_col_blind() + tm_scale_bar()