Using group_by to one-line plot multiple boxplots does not set the x axis labels correctly

[lianos]

First: thank you for the package, I've only just encountered it, but it looks like it has a lot of promise.

It seems like in order to draw multiple box plots on the x-axis w/ out writing e_boxplot(...) for every value over x, we need to use group_by()

So, taking the example from the e_boxplot() man page which has data in a wide format, we have:

library(dplyr)
library(echarts4r)

df <- data.frame(
  x = c(1:10, 25),
  y = c(1:10, -6))

df %>%
  e_charts() %>%
  e_boxplot(x, outliers = TRUE) %>% 
  e_boxplot(y, outliers = TRUE)

If we had multiple categories we wanted to create boxplots for across the x-axis, that can get tedious, so it seems we need to rather convert the data into a long format, and use group_by, like so:

dfl <- data.frame(
  var = rep(c("x", "y"), each = nrow(df)),
  value = c(df$x, df$y))

dfl %>% 
  group_by(var) %>% 
  e_charts() %>% 
  e_boxplot(value)

1. Just want to clarify that this is the suggested way to do that(?)
2. The issue is that when we plot the data in this way, the labels on the x-axis or the boxplots are all "value" instead of "x" and "y".

Perhaps there is a way to use the name parameter in the e_boxplot() to address this, but not quite sure how if so.

Thanks!

[swsoyee]

@lianos I have check the source code, and the labels of x-axis may not be set properly when using group_by() function. But you could manually set labels by using e_x_axis() at present.

Documents are here: https://echarts.apache.org/en/option.html#xAxis.data

library(dplyr)
library(echarts4r)

dfl <- data.frame(
  var = rep(c("x", "y"), each = nrow(df)),
  value = c(df$x, df$y))

dfl %>% 
  group_by(var) %>% 
  e_charts() %>% 
  e_boxplot(value) %>%
  e_x_axis(data = c("x", "y"))

[lianos]

I see, thanks for digging around in the codebase! For now, I'd be too nervous to use that approach as I fear I'd be mislabeling the order of the x-axis.

I'm not all that familiar with the codebase just yet, but it seems you chased the code logic down to the .name_it() call from within the e_boxplot_() function?

Poking around, here's a naive way for a quick fix to the codebase that looks to be working on my side, if we replace the .name_it() call on this line in add_.R with the snippet below, it seems to do the trick:

      if (length(e$x$data) > 1 && serie == name) {
        nm <- .name_it(e, serie, names(e$x$data)[i], i)
      } else {
        nm <- .name_it(e, serie, name, i)
      }

Now my original snippet labels the x axes correctly:

dfl %>% 
  group_by(var) %>% 
  e_charts() %>% 
  e_boxplot(value)

However probably won't be so simple when it's done "professionally" :-)

Just having some fun,
thanks!

[note I edited this comment because my first shot at this I posted was very wrong]

[JohnCoene]

Thanks, there is definitely an issue here somewhere, I'll take a look.

[JohnCoene]

Pushed a fix, hope this works.

df <- data.frame(
  x = c(
    rnorm(100),
    runif(100, -5, 10),
    rnorm(100, 10, 3)
  ),
  grp = c(
    rep(LETTERS[1], 100),
    rep(LETTERS[2], 100),
    rep(LETTERS[3], 100)
  )
)

df %>% 
  group_by(grp) %>% 
  e_charts() %>% 
  e_boxplot(x)

[lianos]

Hi John,

Thanks for taking a look.

This does indeed fix the group_by() use case, however the original code that users will find in the Example section of e_boxplot() which has data in a wide format and one e_boxplot() call per column now has null for the x axis label names.

For instance:

df <- data.frame(
  x = c(1:10, 25),
  y = c(1:10, -6))

df %>%
  e_charts() %>%
  e_boxplot(x, outliers = TRUE) %>% 
  e_boxplot(y, outliers = TRUE)

Gets you this:

[JohnCoene]

Thanks for spotting this, just fixed!