add error bars to stacked bar plot in ggplot2 R - Solved

Question

I have tried to draw the error bars on my stacked bars.

I have two data matrix, one is the mean the and the other one is the standard dev.

    > m_mean
                    TP'   TP*    FP
high / 60         5.445 6.162 1.673
high / 120        5.436 5.098 0.942
high / 180        5.060 3.576 0.242
high / 210        4.705 2.709 0.105
medium high / 60  4.915 6.462 2.086
medium high / 120 4.605 5.652 1.178
medium high / 180 3.940 4.219 0.479
medium high / 210 3.440 3.412 0.208
medium / 60       3.690 7.330 2.290
medium / 120      3.021 6.375 1.611
medium / 180      2.634 4.712 0.659
medium / 210      2.134 3.800 0.482
low / 60          1.626 7.060 3.652
low / 120         1.268 5.626 2.464
low / 180         0.960 3.946 1.379
low / 210         0.703 2.942 1.084
    > m_devs
                        TP'      TP*        FP
high / 60         1.7291515 1.738614 1.5577219
high / 120        1.6925628 1.718531 1.0429399
high / 180        1.6664865 1.610844 0.4976783
high / 210        1.5704908 1.434705 0.3347941
medium high / 60  1.7125150 1.785986 1.7854388
medium high / 120 1.6774352 1.762953 1.1303069
medium high / 180 1.5921479 1.666078 0.7484111
medium high / 210 1.5845854 1.490798 0.4847380
medium / 60       1.6830724 1.751619 1.7569824
medium / 120      1.4651641 1.806000 1.2865980
medium / 180      1.4373970 1.672679 0.8375085
medium / 210      1.2765867 1.527198 0.7498255
low / 60          1.2568626 1.793772 2.0845242
low / 120         1.0475081 1.878204 1.5630570
low / 180         0.9794689 1.666692 1.1818442
low / 210         0.7882972 1.518862 1.0059602

I changed for both matrix the format from WIDE to LONG

library(reshape2)
  table2 <- melt(m_mean)
  table3 <- melt(m_devs)
  colnames(table2) <- c("Combination", "Legend", "Mean")
  colnames(table3) <- c("Combination", "Legend", "SD") 
  attach(table2)
  attach(table3)

I have drawn the graph using ggplot2

library(ggplot2)
  gg<-ggplot(table2, aes(x = Combination, y = Mean, fill = Legend)) + 
    geom_bar(stat = "identity")+ scale_fill_grey()+
    theme_light()+

    labs(title = "MAIN TITLE")+
    theme(axis.text.y  = element_text(face="bold", vjust=0.5, size=10),
          axis.text.x  = element_text(face="bold", vjust=1, size=10,angle = 30, hjust = 1))+
    scale_y_continuous(limits=c(0, 14),breaks = seq(0,14,2))

I have tried to draw the error bars following some useful answers for example: https://stackoverflow.com/search?q=stacked+bars+errors+ggplot but I do not understand why I do not draw the error bars with the correct value and the correct position

library(ggplot2)
  gg<-ggplot(table2, aes(x = Combination, y = Mean, fill = Legend)) + 
    geom_bar(stat = "identity")+ scale_fill_grey()+
    theme_light()+

    toString(lunghezza),sep=""))+
    theme(axis.text.y  = element_text(face="bold", vjust=0.5, size=10),
          axis.text.x  = element_text(face="bold", vjust=1, size=10,angle = 30, hjust = 1))+
    scale_y_continuous(limits=c(0, 14),breaks = seq(0,14,2))
  # Define the top and bottom of the errorbars
  gg <- gg + geom_errorbar(aes(ymax = Mean + SD, ymin=Mean - SD),
                          position = "identity", width = 1)

Errorbars looks fine (good y position), but your bars are stacked. Do you really need stacker bars ? If you need them, you might compute new y values (Mean) for each errorbar, e.g. TP' y should be (Mean TP') + (Mean TP*) instead of (Mean TP*). — bVa
– bVa, Commented Sep 27, 2016 at 12:16
I need to show the three parameters (TP', TP* and FP) as stacked bars. I am trying to draw the error bars for each bar, so 3 error bars per column. — Strobila
– Strobila, Commented Sep 27, 2016 at 12:51

bVa · Accepted Answer · 2016-09-27 13:09:22Z

As I commented, you can directly compute new y_position for each errorbar :

library(plyr)
set.seed(1)
df = data.frame(group = sample(LETTERS[1:3], 100,replace = T),
            var1 = rnorm(100,10,3),
            var2 = rnorm(100,20,5),
            var3 = rnorm(100,50,10))

m_mean = ddply(df, "group", summarize, var1 = mean(var1), var2 = mean(var2), var3 = mean(var3))
m_devs = ddply(df, "group", summarize, var1 = sd(var1), var2 = sd(var2), var3 = sd(var3))

table2 = melt(m_mean)
table3 = melt(m_devs)

colnames(table2)[3] = "Mean"
colnames(table3)[3] = "SD"

table4 = merge(table2, table3)

ggplot(table4, aes(group, Mean, fill = variable)) + 
geom_bar(stat = "identity") + 
geom_errorbar(aes(ymax = Mean + SD, ymin=Mean - SD), position = "identity", width = 1)

Here is a similar plot with random data:

And here is a solution, computing y position before plotting:

# solution
table4$y_pos = NA
table4$y_pos[table4$variable == "var1"] = table4$Mean[table4$variable == "var1"]
table4$y_pos[table4$variable == "var2"] = table4$Mean[table4$variable == "var1"] + 
table4$Mean[table4$variable == "var2"]
table4$y_pos[table4$variable == "var3"] = table4$y_pos[table4$variable == "var2"] + 
table4$Mean[table4$variable == "var3"]

ggplot(table4, aes(group, Mean, fill = variable)) + 
geom_bar(stat = "identity") + 
geom_errorbar(aes(ymax = y_pos + SD, ymin = y_pos - SD), position = "identity", width = 1)

Collectives™ on Stack Overflow

add error bars to stacked bar plot in ggplot2 R - Solved

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related