How to generate sub-plots with grouped categorical variable sorted by a numeric variable in ggplot?

Question

I have a dataframe text with count n of word appearing in each file file_num = 1 or 2 or 3. I would like to use ggplot to generate three subplots, one for each value of file_num, with word on the y-axis and the frequency n on x-axis. I want each sub-plot to be sorted according to increasing or decreasing value of n observed withing each file_num. I have tried many different ways to solve this seemingly trivial issue but have not been successful.

Here is dput of my test data:

structure(list(file_num = c("1", "1", "1", "1", "2", "2", "2", 
"2", "2", "3", "3", "3", "3", "3"), word = c("test", "quality", 
"page", "limit", "information", "limit", "test", "instruments", 
"quality", "limit", "test", "effective", "page", "system"), n = c(5, 
35, 55, 75, 20, 30, 40, 60, 70, 101, 201, 301, 401, 501)), class = c("spec_tbl_df", 
"tbl_df", "tbl", "data.frame"), row.names = c(NA, -14L), spec = structure(list(
    cols = list(file_num = structure(list(), class = c("collector_character", 
    "collector")), word = structure(list(), class = c("collector_character", 
    "collector")), n = structure(list(), class = c("collector_double", 
    "collector"))), default = structure(list(), class = c("collector_guess", 
    "collector")), skip = 1), class = "col_spec"))

Here is what I have tried:

library(tidytext)
library(stringr)
library(pdftools)
library(dplyr)
library(purrr)
library(ggplot2)
library(forcats)
text %>% group_by(file_num) %>% arrange(file_num, desc(n)) %>%
    ggplot(.,aes(factor(word,levels = unique(word)), n, fill = file_num)) + 
    geom_bar(stat = "identity", position = "dodge") +
    scale_x_discrete("Word") +
    scale_y_continuous("n")  + coord_flip() +
    facet_grid(rows = vars(file_num), scales = "free")

Here is the plot that is generated using the above code on dataframe text created using the dput data. It shows the desired result (word sorted with increasing value of n) for file_num = 1, but not for file_num = 2 or 3:

Does this answer your question? [How to order data by value within ggplot facets](https://stackoverflow.com/questions/52214071/how-to-order-data-by-value-within-ggplot-facets) — tjebo, May 02 '20 at 14:15
Instead of all those library calls, a single library call to library(tidyverse) would also do the deal :) — tjebo, May 02 '20 at 14:17
This might also be of use - https://github.com/dgrtwo/drlib/blob/master/R/reorder_within.R — hfshr, May 02 '20 at 14:20
@Tjebo thanks for pointing in the right direction after a lot of wrangling I have a working solution. I will post below. — sachin2014, May 02 '20 at 18:09

score 1 · Answer 1 · answered May 02 '20 at 18:15

Thanks to @Tjebo for pointing me in the right direction. Here is a working solution that is based on ggplot. It does require one to save the modified dataframe text before using it in ggplot.

Let me know if there is a way to directly pipe the modified dataframe into ggplot

text1 <- text %>% ungroup %>% arrange(file_num, n) %>%
            mutate(order = row_number()) # create variable order 

ggplot(text1,aes(order, n, fill = file_num)) + 
    geom_bar(stat = "identity", show.legend = FALSE) +
    scale_x_continuous(
        breaks = text1$order,
        labels = text1$word,
        expand = c(0,0),
        xlab("Word")) +
    facet_grid(file_num ~ ., scales = "free") +
    coord_flip()

Output plot:

KoenV · Accepted Answer · 2020-05-03T10:10:26.473

0

You could achieve this "ordered per facet" quite simply with the ggcharts package, using the following code on your data:

library(ggcharts)
bar_chart(data = text, x = word, y = n, 
  fill = file_num,
  facet = file_num,
  horizontal = TRUE
)

This yields the following graph:

Please, let me know whether this is what you want.

Update:

The object created by bar_chart is of class ggplot, as can be seen below:

class(chart)
[1] "gg"     "ggplot"

This means that one can use the ggplot2 functions to alter the graph, e.g.:

chart + 
  guides(fill=FALSE) +      ## remove legend 
  ggtitle("My new title") + ## add title
  theme_linedraw() +
  theme(strip.background = element_rect(colour = "red", size = 2))

yielding the following pic (for illustration only) :

edited May 03 '20 at 10:10

answered May 02 '20 at 14:25

KoenV

4,113
2
23
38

thanks I will accept your answer but I was really looking for a `ggplot` based approach. – sachin2014 May 02 '20 at 18:16
@sachin2014 Actually, if you type `class(chart)`, you will see that it is a `ggplot`-object. This means that you can use the `ggplot2` package to alter it: e.g. `chart + ggtitle("My new title")` – KoenV May 03 '20 at 07:31

How to generate sub-plots with grouped categorical variable sorted by a numeric variable in ggplot?

2 Answers2