# Ages by England & Wales constituency

There is are a lot of opportunities for data visualizations in journalism and R is beginning to get a toehold in this arena

I recently came across a for-loop tutorial from the R for journalists blog and decided to reprocess it using tidyverse packages whilst adding a plot and map

The article uses some GB office of National Statistics data to calculate the estimated mean age of the population of each of the England and Wales parliamentary constituencies

Let’s load all the libraries required and the downloaded data. With a little bit of inspection, the correct parameters can be applied to the readxl package to produce a data.frame

library(tidyverse)
library(plotly)

library(leaflet)
library(sf)
library(htmlwidgets)
library(DT)

library(hansard)
library(mnis)
library(parlitools)

ages  %>%
DT::datatable(class='compact stripe hover row-border order-column',rownames=FALSE,options= list(paging = FALSE, searching = FALSE,info=FALSE))

So we have an ‘untidy’, wide data.frame with most of the columns representing an age range. For instance, Aldershot had an estimated 1509 children between under 1 year in mid-2015

We can use th gather() function from the tidyr package (part of the tidyverse) to get a row for each age/constituency combination.

We can dispense with the all ages column for this purpose( Note the requirement for a backtick as the column includes a space) but want to retain the other two columns

ages_gather <- ages %>%
select(-All Ages) %>%
gather(age,count,-c(PCON11CD,PCON11NM))

glimpse(ages_gather)
## Observations: 52,143
## Variables: 4
## $PCON11CD <chr> "E14000530", "E14000531", "E14000532", "E14000533", "... ##$ PCON11NM <chr> "Aldershot", "Aldridge-Brownhills", "Altrincham and S...
## $age <chr> "0", "0", "0", "0", "0", "0", "0", "0", "0", "0", "0"... ##$ count    <dbl> 1509, 814, 1079, 916, 848, 1164, 1496, 1219, 1682, 15...

We want to get a weighted mean age for every constuency. That will require amending the age column from a character to an integer and adding 0.5 - on the reasonable assumption that the average age is half of the range - and then using dplyr to calculate the values

meanAges <- ages_gather %>%
mutate(age=as.integer(age)+0.5) %>%
group_by(PCON11CD,PCON11NM) %>%
summarize(wtdMean=round(weighted.mean(age,count),1)) %>%
ungroup()

meanAges %>%
select(Constituency=PCON11NM,Av. Age=wtdMean) %>%
DT::datatable(width=400,class='compact stripe hover row-border order-column',rownames=FALSE,options= list(paging = TRUE, searching = TRUE,info=FALSE))

We now have a searchable, sortable table

The mean age of Aldershot is 38.6. This is 0.5 higher than the original article but I’m assuming the author did not add the aforementioned 0.5. He has the same lowest age - Birmingham Ladywood. Christchurch , which is a favourite retirement town, where my mother happened to die, is the oldest

A histogram will provide a swift look at the distribution

meanAges %>%
plot_ly(x=~wtdMean) %>%
layout(title="Histogram of mean ages by UK constituency, 2015",
xaxis=list(title="Average Age"),
yaxis=list(title="Number of constituencies")) %>%  config(displayModeBar = F,showLink = F)
## No trace type specified:
##   Based on info supplied, a 'histogram' trace seems appropriate.
##   Read more about this trace type -> https://plot.ly/r/reference/#histogram

A, more-or-less, normal distribution but with an age-range greater than I would have guessed going into this exercise

Ok lets look at mapping the data. One interesting method on constructing a hexagon-based electoral cartogram is in a [post by Rob Hickman] (https://robwhickman.github.io/2017/05/16/uk-general-election-hexagram-part-1/) but I have gone with the parlitools package from Evan Odell, whose hansard package I have previously utilized in a couple of earlier posts

The code is basically a copy from his vignette but plugging in the age data

west_hex_map <- parlitools::west_hex_map #Base map

# color range
pal = colorNumeric("Oranges", meanAges$wtdMean) west_mean_ages <- left_join(west_hex_map, meanAges, by = c("gss_code"="PCON11CD")) %>% #Joining to base map filter(!is.na(wtdMean)) # rstrict to England and Wales #create infomatic label label_yes <- paste0( "<strong>", west_mean_ages$constituency_name, "</strong>", "</br>",
"Av. Age: ", west_mean_ages\$wtdMean
) %>% lapply(htmltools::HTML)

# create map
leaflet(options=leafletOptions(
dragging = FALSE, zoomControl = FALSE, tap = FALSE,
minZoom = 6, maxZoom = 6, maxBounds = list(list(2.5,-7.75),list(58.25,50.0)),
west_mean_ages) %>%
color = "grey",
weight=0.75,
opacity = 0.5,
fillOpacity = 1,
fillColor = ~pal(wtdMean),
label = label_yes) %>%

addLegend("topright", pal = pal, values = ~wtdMean,
title = "Mean Age",
opacity = 1)  %>%
htmlwidgets::onRender(
"function(x, y) {
var myMap = this;
myMap._container.style['background'] = '#fff';
}")%>%
mapOptions(zoomToLimits = "first")

Hover map for details. This approach neatly highlights that there is a preponderance of younger people living in the middle of the major cities, e.g. London, Birmingham , Leeds. In contrast, coastal regions - attractive to retirees - have much higher average ages