CODEWITHSUNDEEP
rdata-analysis
D. O Sullivan
D. O Sullivan

Reputation: 199

barplot - Grouping x-axis labels without manipulating accompanying bars

I'm doing some basic data analysis on this dataset: https://www.kaggle.com/murderaccountability/homicide-reports

I'm generating a basic barplot using the State names as the x-axis values, and the y-axis values is the percentage of nationwide homicide occurrences (number of entries in the data set divided by the total number of entries) 

barplot(prop.table(table(homicideData.raw$State)),
    main = "Nationwide Homicide % per State",
    ylab = "Accounting % of Nation-wide Homicides",
    las=2)

enter image description here

This is very messy, is there a way of grouping perhaps 5 states together as an x-axis label, without changing the bars?

Let's say the following for example:

x-axis labels: "Alaska - California", "Colorado - Florida", ... (and so on). Each label should then have 5 bars above it.

Upvotes: 1

Views: 261

Answers (1)

Adam Quek
Adam Quek

Reputation: 7163

Here's a solution with ggplot. It's not the simplest as it involves some data-manipulation.

(1) read in the data-set and extract the homicide count/proportion by state:

df <- read.csv("homicide.csv")

library(dplyr)
freq <- with(df, table(State)) %>% data.frame
freq <- freq %>% mutate(prop = Freq/sum(Freq))

(2) find first and last element of each group of 5 states:

hd <- seq(1, nrow(freq), by=5) %>% ceiling
hd <- hd[-length(hd)]
td <- c((hd-1)[-1], nrow(freq))

(3) custom function to make the custom label for each group (e.g. Alb - Clf) and calculate length of each group

abbrevFn <- function(head, tail, state, ...) paste(abbreviate(state[c(head,tail)], ...), collapse = " - ")

intervalFn <- function(head, tail) diff(c(head, tail)) + 1

(4) group the states by replicating custom label by the length for each group

freq$group <- lapply(1:length(hd), function(x) rep(abbrevFn(hd[x], td[x], freq$State, min=3), intervalFn(hd[x], td[x]))) %>% unlist

(5) plot geom_bar based on the customised group, and dodge position by state:

xint <- c((1:length(hd) - .5), (1:length(hd) + .5)) %>% unique

library(ggplot2)
ggplot(freq, aes(group, prop, fill=State)) + 
  geom_bar(stat="identity", position="dodge", width=1) + 
  scale_fill_manual(values=rep("gray80", nrow(freq))) +
  ylab("Accounting % of Nation-wide Homicides") +
  xlab("States") +
  geom_vline(xintercept=xint, linetype="dotted") +
  guides(fill=FALSE) +
  theme_bw()

enter image description here

Upvotes: 2

Related Questions