CODEWITHSUNDEEP
rggplot2
Ben
Ben

Reputation: 21625

Split facet plot into list of plots

Huge fan of facet plots in ggplot2. However, sometimes I have too many subplots and it'd be nice to break them up into a list of plots. For example

df <- data.frame(x=seq(1,24,1), y=seq(1,24,1), z=rep(seq(1,12),each=2))
df
    x  y  z
1   1  1  1
2   2  2  1
3   3  3  2
4   4  4  2
5   5  5  3
.   .  .  .
.   .  .  .

myplot <- ggplot(df,aes(x=x, y=y))+geom_point()+facet_wrap(~z)
myplot

enter image description here

How would I write a function to take the resulting plot and split it into a list of plots? Something along these lines

splitFacet <- function(subsPerPlot){
  # Method to break a single facet plot into a list of facet plots, each with at most `subsPerPlot` subplots

  # code...

  return(listOfPlots)
}

Upvotes: 20

Views: 5437

Answers (4)

Nathan Werth
Nathan Werth

Reputation: 5263

This is similar to Moody_Muddskipper's answer, but works with any type of faceting (facet_grid or facet_wrap), handles arbitrary expressions in facets, and doesn't draw facet strip bars.

library(rlang)
library(ggplot2)

split_facets <- function(x) {
  facet_expr <- unlist(x[["facet"]][["params"]][c("cols", "rows", "facets")])
  facet_levels <- lapply(facet_expr, rlang::eval_tidy, data = x[["data"]])
  facet_id <- do.call(interaction, facet_levels)
  panel_data <- split(x[["data"]], facet_id)
  plots <- vector("list", length(panel_data))
  for (ii in seq_along(plots)) {
    plots[[ii]] <- x
    plots[[ii]][["data"]] <- panel_data[[ii]]
    plots[[ii]][["facet"]] <- facet_null()
  }
  plots
}

split_facets(ggplot(df,aes(x=x, y=y))+geom_point()+facet_wrap(~z))
split_facets(ggplot(df,aes(x=x, y=y))+geom_point()+facet_grid(z %% 2 ~ z %% 5))

It uses rlang::eval_tidy to evaluate the facet expressions, combines them into a single categorical factor, then uses that to split the data. It also "suppresses" each subplot's faceting part by replacing it with facet_null().

Upvotes: 3

moodymudskipper
moodymudskipper

Reputation: 47310

Split plot into individual plots

We build a function along these steps :

  1. We go through the structure of the object to get the names of the variables used for faceting (here 'z').
  2. We overwrite the facet element of our plot object with the one from the empty ggplot object (so if we print it at this stage facets are gone).
  3. We extract the data and split it along the variables we identified in 1st step.
  4. We overwrite the original data with each subset (12 times here) and store all outputs in a list.

code

splitFacet <- function(x){
  facet_vars <- names(x$facet$params$facets)         # 1
  x$facet    <- ggplot2::ggplot()$facet              # 2
  datasets   <- split(x$data, x$data[facet_vars])    # 3
  new_plots  <- lapply(datasets,function(new_data) { # 4
    x$data <- new_data
    x})
}    

new_plots <- splitFacet(myplot)
length(new_plots) # [1] 12
new_plots[[3]]    # 3rd plot

Split plot into faceted plots of n subplots max

If we want to keep the facets but have less plots by facet we can skip step 2, and rework our split instead so it includes several values of the variables used for faceting.

Rather than making a separate function we'll generalize the 1st, n is the number of facets you get by plot.

n = NULL means you get the previous output, which is slightly different from n = 1 (one facet by plot).

splitFacet <- function(x, n = NULL){
  facet_vars <- names(x$facet$params$facets)               # 1
  if(is.null(n)){
    x$facet  <- ggplot2::ggplot()$facet                    # 2a
    datasets <- split(x$data, x$data[facet_vars])          # 3a
  } else {
    inter0 <- interaction(x$data[facet_vars], drop = TRUE) # 2b
    inter  <- ceiling(as.numeric(inter0)/n)
    datasets <- split(x$data, inter)                       # 3b
  }
  new_plots  <- lapply(datasets,function(new_data) {       # 4
    x$data <- new_data
    x})
} 

new_plots2 <- splitFacet(myplot,4)
length(new_plots2) # [1] 3
new_plots2[[2]]

This might come in handy too :

unfacet <- function(x){
  x$facet <- ggplot2::ggplot()$facet
  x
}

The tidy way

If the code is available, no need to go through all this trouble, we can split the data before feeding it to ggplot :

library(tidyverse)
myplots3 <-
  df %>% 
  split(ceiling(group_indices(.,z)/n_facets)) %>% 
  map(~ggplot(.,aes(x =x, y=y))+geom_point()+facet_wrap(~z))

myplots3[[3]]

Upvotes: 20

luvdata
luvdata

Reputation: 9

Posting this for anyone wanting to use ggplus. ggplus will work with later versions of R, but you need to install it using the developer's directions, i.e.

 devtools::install_github("guiastrennec/ggplus")

I ran into the same issue when trying to install it using RStudio, then realized that it's just not one of the "standard packages." I'm using 3.4.4.

Upvotes: 0

Pete900
Pete900

Reputation: 2166

While I was looking for a solution for this I can across ggplus. Specifically the function facet_multiple:

https://github.com/guiastrennec/ggplus

It lets you split a facet over a number of pages by specifying the amount of plots you want per page. In your example it would be:

library(ggplus)

df <- data.frame(x=seq(1,24,1), y=seq(1,24,1), z=rep(seq(1,12),each=2))

myplot <- ggplot(df,aes(x=x, y=y))+geom_point()

facet_multiple(plot = myplot, facets = 'z', ncol = 2, nrow = 2)

Is this the sort of thing you need? It worked a treat for me.

Upvotes: 5

Related Questions