Grouped plot with consideration of the row order of the given data-CodePudding

Suppose the following data set is given:

df <- data.frame(x1 = seq(1,8,1),
                 x2 = c("G1","G2","G1","G2",
                        "G1","G2","G1","G2"))

I want to create a grouped plot, for example a grouped bar chart. But I want to use the order of the rows of the dataset for the plot, i.e. instead of combining all values of group 1 and group 2 and then plotting them, I want the bar plot to have a kind of alternating structure in which first the value of group 1 is displayed in red as specified in the dataset, then blue for group 2, then red again for group 1, and so on. Is this possible? I have added a picture of what I envision.

CodePudding user response：

Maybe you want something like this, which you modify the plot by changing the values using ggplot_build so that you have the same order as your dataframe like this:

df <- data.frame(x1 = seq(1,8,1),
                 x2 = c("G1","G2","G1","G2",
                        "G1","G2","G1","G2"))

df$group <- ""
library(ggplot2)
p <- ggplot(df, aes(x = group, y = x1, fill = x2))   
  geom_bar(position="stack", stat="identity")  
  labs(x = "", y = "")  
  coord_flip()

q <- ggplot_build(p)

q$data[[1]]$ymin <- c(0, 1, 3, 6, 10, 15, 21, 28)
q$data[[1]]$ymax <- c(1, 3, 6, 10, 15, 21, 28, 36)

q <- ggplot_gtable(q)
plot(q)

^{Created on 2022-08-25 with reprex v2.0.2}

CodePudding user response：

It seems like these groups have values for specific time points? You could add an additional variable for each time point and then generate the plot as followed:

library(tidyverse)
df <- data.frame(x1 = seq(1,8,1),
                 x2 = c("G1","G2","G1","G2",
                        "G1","G2","G1","G2"),
                 time = rep(c(paste("t", c(1:4), sep = "")), each = 2)) 

df %>%
  ggplot(aes(y = x1, x = time, fill = x2)) 
  geom_bar(stat = "identity", position = "dodge2")

^{Created on 2022-08-25 with reprex v2.0.2}

CodePudding user response：

Maybe you could replace the part

q$data[[1]]$ymin <- c(0, 1, 3, 6, 10, 15, 21, 28)
q$data[[1]]$ymax <- c(1, 3, 6, 10, 15, 21, 28, 36)

with the following part:

y_min <- rep(0,length(df$x1))
y_max <-  rep(0,length(df$x1))
for(k in 2:length(df$x1)){
y_min[k] <- sum(df$x1[1:(k-1)])  
}
y_max[1:(length(y_max)-1)] <- y_min[2:length(y_min)]
y_max[length(y_max)] <- max(y_min) max(df$x1)
q$data[[1]]$ymin <- y_min
q$data[[1]]$ymax <- y_max

For some level of generality.