Creating a grouped barchart in R, but with unique x values in each group?

Question

Hopefully this isn't too obvious of an issue to solve or hasn't been asked yet.

I have data that is a much longer version of this: | Publisher| Name | Sales| | -------- | -----|----- | | Nintendo | Aa | 30 | | Sony | Ab | 20 | | EA | Ac | 20 | | Nintendo | Ad | 15 | | Sony | Ae | 10 | | Sony | Af | 20 | | EA | Ag | 30 |

And I would like to make a bar chart similar to this, except I am searching for top five games for each of the top ten publishers.

This is the code I tried, but I find facet_wrap isn't what I want:

pubsales <- df%>%group_by(Publisher)%>%
  summarise(pubttlsale=sum(Sales))%>%
  top_n(10)%>%
  arrange(desc(pubttlsale))
df2 <- df%>%group_by(Publisher)%>%filter(Publisher%in%pubsales$Publisher)%>%
  summarise(pubttlsale=sum(Sales),Name,Publisher,Sales)%>%
  arrange(desc(pubttlsale))%>%
  top_n(5)
df2%>%ggplot(aes(x=Name,y=Sales))+
  geom_bar(stat="identity")+
  facet_grid(~Publisher)+
  coord_flip()+
  labs(x="Game",y="Total Sales (millions of copies)")

Thank you for reading. Please tell me if I need to clarify more or provide more sample data (it's actually this set from Kaggle).

Answer 1

A simple approach to achieve your desired result would be to make use of facet_wrap(~Publisher, scales = "free_y", ncol = 1, strip.position = "right") which will put all publishers in one column, shows only the games per publisher and puts the script text on the right.

Note: To make the example both minimal and reproducible I used only the top 3 publishers and the top 4 games and added the data via dput(df2) See how to make a minimal reproducible example.

library(ggplot2)

ggplot(df2, aes(x = Name, y = Sales)) +
  geom_bar(stat = "identity") +
  facet_wrap(~Publisher, scales = "free_y", ncol = 1, strip.position = "right") +
  coord_flip() +
  labs(x = "Game", y = "Total Sales (millions of copies)")

To get more closer to the chart you posted as a link you could reorder the Publishers by sales and make use of tidy text::reorder_within + tidy text::scales_y_reorderd to order Name by Sales per Publisher, make use of geom_text to put the publisher's name on the plot which allows to get rid of the strip texts and some additional styling.

Note: To simplify the code I make use of geom_col which is short for geom_bar(stat = "identity") and switched the role of x and y to get rid of the coord_flip:

library(dplyr)
library(tidytext)

df2 <- df2 %>%
  mutate(
    Publisher = reorder(Publisher, -Sales),
    Name = tidytext::reorder_within(Name, Sales, Publisher, fun = sum)
  ) %>% 
  group_by(Publisher) %>% 
  mutate(y_label = Name[Name == last(Name)],
         label = ifelse(Name %in% y_label, as.character(Publisher), ""))

ggplot(df2, aes(x = Sales, y = Name, fill = Publisher)) +
  geom_col() +
  geom_text(aes(x = max(Sales), y = y_label, label = label), hjust = 1) +
  tidytext::scale_y_reordered() +
  scale_x_continuous(expand = expansion(mult = c(0, .05))) +
  facet_wrap(~Publisher, scales = "free_y", ncol = 1) +
  labs(x = "Total Sales (millions of copies)", y = "Game") +
  guides(fill = "none") +
  theme_minimal() +
  theme(
    strip.text = element_blank(),
    panel.grid.major.y = element_blank(),
    panel.grid.minor = element_blank()
  )

DATA

df2 <- structure(list(Publisher = c(
  "Nintendo", "Nintendo", "Nintendo",
  "Nintendo", "Electronic Arts", "Electronic Arts", "Electronic Arts",
  "Electronic Arts", "Activision", "Activision", "Activision",
  "Activision"
), pubttlsale = c(
  1788.81, 1788.81, 1788.81, 1788.81,
  1116.96, 1116.96, 1116.96, 1116.96, 731.16, 731.16, 731.16, 731.16
), Name = c(
  "Wii Sports", "Super Mario Bros.", "Mario Kart Wii",
  "Wii Sports Resort", "FIFA 16", "FIFA Soccer 13", "The Sims 3",
  "Star Wars Battlefront (2015)", "Call of Duty: Modern Warfare 3",
  "Call of Duty: Black Ops 3", "Call of Duty: Black Ops", "Call of Duty: Black Ops II"
), Sales = c(
  82.53, 40.24, 35.52, 32.77, 8.57, 8.16, 8.01, 7.98,
  14.73, 14.63, 14.61, 13.79
)), class = c("tbl_df", "tbl", "data.frame"), row.names = c(NA, -12L))