Horizontal bar chart split by categories

Question

I have the following data frame

A <- c(2020, 2020, 2020, 2021, 2021, 2021)
B <- c("category 1", "category 2", "category 3", "category 1", "category 2", "category 3")
C <- c(7, 10, 5, 11, 8, 7)

data <- data.frame(A, B, C)
names(data) <- c("yeat", "categories", "value")

How can I plot a horizontal bar chart, that will look like this (i.e., the categories are sorted by their values for each year):

Answer 1

How about using r-plotly as the plotting library?

library(plotly)
library(dplyr)

#your sample data
A <- c(2020, 2020, 2020, 2021, 2021, 2021)
B <- c("category 1", "category 2", "category 3", "category 1", "category 2", "category 3")
C <- c(7, 10, 5, 11, 8, 7)    
data <- data.frame(A, B, C)
names(data) <- c("year", "categories", "value")

#subplotting function
sub_plotting <- . %>% 
        #create horizontal bar chart with custom colors and size
        plot_ly(x = ~value, y = ~categories, 
                type="bar", orientation="horizontal", 
                color = ~categories, colors = c('#3C79A7', '#8ECBE9', '#FB4C58'),
                width = 1500, height = 300) %>%
        #add title to subplot
        add_annotations(text = ~year,
                x = 0.5, y = 0.98,
                xref = "paper", yref = "paper",
                xanchor = "center", yanchor = "bottom",
                showarrow = FALSE,
                font = list(size = 25)) %>%  
        #and sort values   
        layout(yaxis=list('categoryorder'='total ascending'), showlegend=FALSE) 

#apply this to grouped data
fig <- data %>%
        group_by(year) %>%      
        do(p = sub_plotting(.)) %>%     
        subplot()

print(fig) 

Sample output:

The definitions of the custom colors are shamelessly stolen from this answer.

Answer 2

Use barplot in an lapply. split the data by year and plot the values according to their order. We also give the bars col=ors according to this scheme as well as labels using mtext.

Colors and labels may be conveniently stored in a matrix m with row names by categories that we can easily match later.

(m <- matrix(c('#3C79A7', '#8ECBE9', '#FB4C58',
              "Category 1", "Category 2", "Category 3"), 3, 2,
            dimnames=list(sort(unique(data$categories)), NULL)))
#                    [,1]      [,2]        
# category 1 "#3C79A7" "Category 1"
# category 2 "#8ECBE9" "Category 2"
# category 3 "#FB4C58" "Category 3"

S <- split(data, data$year)
    
op <- par(mfrow=c(1, 2), mar=c(2, 6, 4, 2))  ## par: 2 plots, expand margins

lapply(names(S), \(x) {
  s <- S[[x]]
  u <- order(s$value)
  barplot(s$value[u], horiz=TRUE, col=m[s$categories[u], 1], border=NA,
          main=x[1], xaxt='n', yaxt='n', ylab='')
  mtext(m[s$categories[u], 2], 2, 1, at=seq_len(nrow(s)), las=2)
})

par(op)  ## reset pars

---

Data:

data <- structure(list(year = c(2020, 2020, 2020, 2021, 2021, 2021), 
    categories = c("category 1", "category 2", "category 3", 
    "category 1", "category 2", "category 3"), value = c(7, 10, 
    5, 11, 8, 7)), class = "data.frame", row.names = c(NA, -6L
))

Answer 3

You can use this code:

library(tidyverse)
library(forcats)
library(patchwork)
A <- c(2020, 2020, 2020, 2021, 2021, 2021)
B <- c("category 1", "category 2", "category 3", "category 1", "category 2", "category 3")
C <- c(7, 10, 5, 11, 8, 7)

data <- data.frame(A, B, C)
names(data) <- c("yeat", "categories", "value")

p1 <- data %>%
  filter(yeat == 2020) %>%
  mutate(categories = fct_reorder(categories, value)) %>%
  ggplot(aes(x = categories, y = value)) +
  geom_col() +
  xlab("Categories") +
  ylab("") +
  ggtitle("2020") +
  coord_flip() +
  theme(legend.position = "none")

p2 <- data %>%
  filter(yeat == 2021) %>%
  mutate(categories = fct_reorder(categories, value)) %>%
  ggplot(aes(x = categories, y = value)) +
  geom_col() +
  xlab("Categories") +
  ylab("") +
  ggtitle("2021") +
  coord_flip() +
  theme(legend.position = "none")

p1 + p2

Output: