R Pie Donut chart with facet functionality-CodePudding

The

Wondering how to construct Pie Donut chart with facet functionality. My attempt is below:

ggPieDonut(data = acs, mapping = aes(pies = Dx, donuts = smoking), interactive = TRUE) facet_wrap(facets = vars(sex)) NULL

CodePudding user response：

It seems @jpiversen and I had the same idea about this, so I initially didn’t post my answer after seeing theirs. However, by request of the OP, and in case it proves useful in some way, here’s my take on this:

I think the short answer is: you can’t.

There’s no option for facetting in the function call, and the resulting ggplot2 object has had it’s data modified enough to not have access to the original columns anymore, making ggplot facetting afterwards impossible.

However, you can get in the vicinity of facetting by constructing the plots separately for each level of the variable you wanted to facet by, and then assembling the plots together.

Here’s how I’d approach that:

library(tidyverse) library(ggiraphExtra) library(moonBook) make_plot <- function(data) { ggPieDonut(data = data, mapping = aes(pies = Dx, donuts = smoking)) } plots <- by(acs, acs$sex, make_plot) combined_plots <- patchwork::wrap_plots(plots) patchwork::plot_annotation(tag_levels = list(names(plots))) combined_plots

I don’t know how to do the combination step if you specify interactive = TRUE, but as a workaround some approximation of interactivity could be also reached with a simple girafe() call on the combined plots:

ggiraph::girafe(code = print(combined_plots))

CodePudding user response：

The problem

The code in your attempt doesn't work because when interactive = TRUE, ggPieDonut() doesn't return a ggplot, but a htmlwidget:

ggPieDonut( data = acs, mapping = aes(pies = Dx, donuts = smoking), interactive = TRUE ) %>% class() #> [1] "girafe" "htmlwidget"

And facet_wrap() only works with ggplots.

If you change to interactive = FALSE you get another problem:

ggPieDonut( data = acs, mapping = aes(pies = Dx, donuts = smoking), interactive = FALSE ) facet_wrap(~sex) #> Error in `combine_vars()`: #> ! At least one layer must contain all faceting variables: `sex`.

The geoms doesn't contain both values of sex, so facet_wrap() doesn't know how to facet on it.

Possible workaround

A solution is to create two plots on different subsets of the data, and use patchwork to combine the two plots:

library(patchwork) p1 <- acs %>% filter(sex == "Male") %>% ggPieDonut(mapping = aes(pies = Dx, donuts = smoking), interactive = FALSE) labs(title = "Male") p2 <- acs %>% filter(sex == "Female") %>% ggPieDonut(mapping = aes(pies = Dx, donuts = smoking), interactive = FALSE) labs(title = "Female") p1 p2

Output:

Update

As @MikkoMarttila suggested, it might be better to create this as a function. If I were to reuse the function, I would probably write it like this:

make_faceted_plot <- function(data, pie, donut, facet_by) { data %>% dplyr::pull( {{facet_by}} ) %>% unique() %>% purrr::map( ~ data %>% dplyr::filter( {{facet_by}} == .x) %>% ggiraphExtra::ggPieDonut( ggplot2::aes(pies = {{pie}}, donuts = {{donut}}), interactive = FALSE ) ggplot2::labs(title = .x) ) %>% patchwork::wrap_plots() }

This can then be used to facet on however many categories we want, and on any dataset, for example:

library(patchwork) library(dplyr) # Expandable example data df <- data.frame( eyes = sample(c("Blue", "Bown", "Green"), size = 100, replace = TRUE), hair = sample(c("blonde", "brunette", "raven"), size = 100, replace = TRUE), sex = sample(c("male", "female"), size = 100, replace = TRUE) ) df %>% make_faceted_plot( pie = eyes, donut = sex, facet_by = hair )

Again, as suggested by @MikkoMarttila, this can be piped into ggiraph::girafe(code = print(.)) to add some interactivity.