Indicating direction of change in a dumbbell plot using arrows and colors in ggplot2-CodePudding

I have the following dataset:

Output of dput(prob_metrics) is:

structure(list(round = c("baseline", "baseline", "baseline", 
"baseline", "baseline", "baseline", "baseline", "baseline", "baseline", 
"endline", "endline", "endline", "endline", "endline", "endline", 
"endline", "endline", "endline"), metric = structure(c(1L, 2L, 
3L, 4L, 5L, 6L, 7L, 8L, 9L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L
), levels = c("Kesehatan yang buruk", "Konflik keluarga", "Tingkatan penghasilan", 
"Kekerasan dalam Rumah Tangga", "Kenakalan", "Kurangnya kepercayaan atau respek", 
"Rendahnya Kerjasama Masyarakat", "Kurangnya keamanan", "Kekurangan ada pendampingan"
), class = "factor"), percentage = c(82.1, 82.6, 77.7, 80.6, 
78.1, 81.2, 78.2, 66.2, 75.9, 62.3, 70.9, 76.2, 82.8, 98.4, 96.5, 
95.8, 95.1, 79), bump_percentage = c(80.1, 80.6, 75.7, 78.6, 
76.1, 79.2, 76.2, 64.2, 73.9, 64.3, 72.9, 78.2, 84.8, 100.4, 
98.5, 97.8, 97.1, 81), midpoint = c(72.2, 76.75, 76.95, 81.7, 
88.25, 88.85, 87, 80.65, 77.45, 72.2, 76.75, 76.95, 81.7, 88.25, 
88.85, 87, 80.65, 77.45)), row.names = c(NA, -18L), class = c("tbl_df", 
"tbl", "data.frame"))

I have created this plot

using this code:

mh_theme <-theme(panel.background = element_rect(fill = "#F2E6CE"),
                 plot.background = element_rect(fill = "#F2E6CE"),
                 legend.background = element_rect(fill = "#F2E6CE"),
                 panel.grid.major = element_line(color = "#DCD1BC"),
                 legend.title = element_blank())
ggplot(prob_metrics, aes(x=percentage, y=metric))   
  geom_line(color="gray", size=1.75)  
  geom_segment(aes(x = midpoint, xend = midpoint   1, yend = metric), 
               arrow = arrow(angle = 30, length = unit(.15, "inches"), type = "closed"), 
               color = "gray", )  
  geom_point(size=2)   
  geom_text(aes(label=glue("{percentage}%"), x=percentage), 
            color = "dimgray", size=3, nudge_y = .2)  
  theme_fivethirtyeight()   mh_theme   xlim(50,102)   
  labs(title = str_wrap("Persentase anak dengan status gizi buruk yang sehat", 38))  
  scale_y_discrete(labels = function(x) str_wrap(x, width = 20))

Now here is my problem: up until this plot, the endline value has always been bigger than the baseline value of every single variable in my analysis. Now, I have some values that decrease from baseline to endline. To visualize this, I want to do two things that are giving me a surprising amount of trouble:

I want to color the baseline points with the hex code "#3C4C56" and the endline with "#DB5628". I know how to color the points by round in theory, and I've done it successfully here using Rcolorbrewer, but I can only get it to work using their remade palettes. I tried:

base_end_colors <- as.character(c("baseline"="#3C4C56", "endline" = "#DB5628"))
base_end_scale <- scale_color_manual(name="round", values = base_end_colors)

and then my_plot base_end_scale but that didn't work. So problem #1 is coloring the points those two colors based on the round variable.

I would also like the arrows to point from baseline to endline, rather than simply pointing to the right, like they do right now. My geom_segment trick for plotting the arrows was, I knew when I coded it, not particularly strong or flexible. Now its met its limit. How could I most simply get these arrows to point from baseline to endline.

CodePudding user response：

Your color issue is easily fixed by mapping on the color aes and setting your desired colors via scale_color_manual. To fix the direction of your arrows you could first filter the data for the baseline values in geom_segement (Otherwise your are drawing the arrows two times) and use and ifelse or the sign function as I do below to "set" the direction of the segment and the arrow.

library(ggplot2)
library(ggthemes)
library(stringr)
library(glue)

ggplot(prob_metrics, aes(x=percentage, y=metric))   
  geom_line(color="gray", size=1.75)  
  geom_segment(data = ~subset(.x, round == "baseline"), 
               aes(
    x = midpoint,
    xend = midpoint   .5 * sign(midpoint - percentage), 
    yend = metric), 
    arrow = arrow(angle = 30, length = unit(.15, "inches"), type = "closed"), 
    color = "gray")  
  geom_point(aes(color = round), size= 2)   
  geom_text(aes(label=glue("{percentage}%"), x=percentage), 
            color = "dimgray", size=3, nudge_y = .2)  
  theme_fivethirtyeight()   mh_theme   xlim(50,102)   
  labs(title = str_wrap("Persentase anak dengan status gizi buruk yang sehat", 38))  
  scale_y_discrete(labels = function(x) str_wrap(x, width = 20))  
  scale_color_manual(values = c(baseline = "#3C4C56", endline = "#DB5628"))  
  theme(legend.key = element_rect(fill = NA))