For these boxplots with custom y quantiles and centered on the medians x_q50, how to set their width based on given x quantiles (e.g. based on x_min and x_max, or x_q025 and x_q975...), while having the possibility to transform the x scale?
I tried this interesting suggestion but without success.
Thanks for help.
Graph attempt (failed):
Desired graph:
For example below (built manually with PowerPoint), the width of the boxplots is defined by the 5th and 95th quantiles of x (yellow x_q05 and x_q95 columns), with the medians being x_q50. For the last group, x_q05 = 82 and x_q95 = 432.8, and x_q50 = 132.
Data:
dat <-
structure(list(grp = structure(1:9, levels = c("0", "1", "2",
"3", "4", "5", "6", "7", "8"), class = "factor"), nb = c(4364L,
4363L, 4363L, 4363L, 4363L, 4363L, 4363L, 4363L, 4363L), x_min = c(0,
5, 8, 11, 15, 20, 28, 43, 79), x_q025 = c(`2.5%` = 0, `2.5%` = 5,
`2.5%` = 8, `2.5%` = 11, `2.5%` = 15, `2.5%` = 20, `2.5%` = 28,
`2.5%` = 43, `2.5%` = 80), x_q05 = c(`5%` = 0, `5%` = 5, `5%` = 8,
`5%` = 11, `5%` = 15, `5%` = 20, `5%` = 28, `5%` = 44, `5%` = 82
), x_q50 = c(3, 7, 9, 13, 17, 23, 34, 56, 132), x_q95 = c(`95%` = 5,
`95%` = 8, `95%` = 11, `95%` = 14, `95%` = 19, `95%` = 27, `95%` = 42,
`95%` = 76, `95%` = 432.8), x_q975 = c(`97.5%` = 5,
`97.5%` = 8, `97.5%` = 11, `97.5%` = 15, `97.5%` = 20, `97.5%` = 28,
`97.5%` = 42, `97.5%` = 77, `97.5%` = 576.8), x_max = c(5,
8, 11, 15, 20, 28, 43, 79, 1932), y_q025 = c(`2.5%` = 8.22375,
`2.5%` = 8.9525, `2.5%` = 9.031, `2.5%` = 9.96, `2.5%` = 10.1,
`2.5%` = 10.205, `2.5%` = 10.9, `2.5%` = 11.405, `2.5%` = 12.4
), y_q05 = c(`5%` = 10.5, `5%` = 11.8, `5%` = 12.01, `5%` = 12.31,
`5%` = 12.5, `5%` = 13.1, `5%` = 13.21, `5%` = 13.6, `5%` = 14.8
), y_q50 = c(28.9, 30, 29.7, 29.8, 29.5, 30.1, 30.3, 30.9, 32.8
), y_q95 = c(`95%` = 137, `95%` = 113, `95%` = 109, `95%` = 104,
`95%` = 105.9, `95%` = 101, `95%` = 98.58, `95%` = 90.39,
`95%` = 98.03), y_q975 = c(`97.5%` = 185.925, `97.5%` = 158,
`97.5%` = 152, `97.5%` = 152, `97.5%` = 155, `97.5%` = 139, `97.5%` = 136,
`97.5%` = 128, `97.5%` = 132.95), varx = c(5, 3, 3, 4, 5, 8,
15, 36, 1853)), row.names = c(NA, -9L), class = c("tbl_df", "tbl",
"data.frame"))
Code attempt:
library(tidyverse)
library(scales)
ggplot(dat, aes(x = x_q50)) +
geom_boxplot(
aes(
ymin=y_q025,
lower=y_q05,
middle=y_q50,
upper=y_q95,
ymax=y_q975,
fill=grp),
stat="identity",
varwidth = TRUE) +
scale_x_continuous(limits=c(0, 2000),
breaks = c(0,10,20,30,40,50,100,150,200,
250,500,750,1000,1500,2000),
labels = c(0,10,20,30,40,50,100,150,200,
250,500,750,1000,1500,2000),
trans = modulus_trans(0.2)) +
scale_fill_manual(values = c("#FEE186", "#FEB24C", "#FD9E45",
"#FD8A3E", "#FC7637", "#FC6230",
"#FC4E2A","#E33E29", "#CA2E28"))
