Force facet wrapped plots to center at a common point, namely 0-CodePudding

I'm sure there's a more elegant way to get to this plot, but here goes with some of the data I have to work with

score <- c(1, 2, 3, 4, 5, "X")
group1 <- c(0,0,5,-0.2,-4.9,0)
group2 <- c(0.1,0,-1.2,0.4,0.6,0.1)
group3 <- c(0.1,0,3.4,2.9,-6.4,0)
group4 <-c(0,0,-0.9,-0.3,1.3,0)

data <- data.frame(score=as.factor(score), group1=group1, group2=group2, group3=group3, group4=group4)

data_long <- pivot_longer(data, c(group1, group2, group3, group4))
sig <- rep(0, 24)
sig[c(3, 9, 11, 15, 17, 19)] <- 1
sig <- as.factor(sig)
data_long <- cbind(data_long, sig)


bars <- data_long %>% filter(score != 2,
                             score != "X") %>%
  ggplot(aes(x = name, y=value, fill=sig))  
  geom_bar(stat="identity")  
  coord_flip()  
  scale_fill_manual(values=c("grey", "black"))  
  facet_wrap(~rating, scales = "free_x")  
  theme(legend.position = "none")

print(bars)

This gives the following plot

What this represents are deviations from expected value for each of the score levels for the four different groups. (it's a "feature" of the problem that we don't care about 2's and X's). Those measures with a 1 in the sig column are statistically significant.

What I'd like to do is have each of the facets centered at 0.

I tried following methods found here but couldn't get things to work...in particular I kept getting Error: Discrete value supplied to continuous scale when attempting the method that made use of geom_blank()

CodePudding user response：

You can provide a function to the limits of a scale. In the example below, note that I've used score instead of rating for facets, and circumvented the coord_flip() by swapping the x- and y-variables. The function below is in the so called 'lambda syntax', but you can use regular function(x) c(-1, 1) * max(abs(x)) too.

library(tidyverse)

# Data prep as in original question
# data_long <- ...


bars <- data_long %>% filter(score != 2,
                             score != "X") %>%
  ggplot(aes(x = value, y=name, fill=sig))  
  geom_bar(stat="identity")  
  scale_fill_manual(values=c("grey", "black"))  
  scale_x_continuous(
    limits = ~ c(-1, 1) * max(abs(.x))
  )  
  facet_wrap(~score, scales = "free_x")  
  theme(legend.position = "none") 

bars