CODEWITHSUNDEEP
rggplot2
Johan Dingler
Johan Dingler

Reputation: 61

Plotting standard error bars

I have a long format dataset with 3 variables. Im plotting two of the variables and faceting by the other one, using ggplot2. I'd like to plot the standard error bars of the observations from each facet too, but I've got no idea how. Anyone knows?

Here´s a picture of what i've got. I'd like to have the standard error bars on each facet. Thanks!!

Edit: here's some example data and the plot.

data <- data.frame(rep(c("1","2","3","4","5","6","7","8","9","10",
                         "11","12","13","14","15","16","17","18","19","20",
                         "21","22","23","24","25","26","27","28","29","30",
                         "31","32"), 2),
                   rep(c("a","b","c","d","e","f","g","h","i","j","k","l"), 32),
                   rnorm(n = 384))

colnames(data) <- c("estado","sector","VA")

ggplot(data, aes(x = estado, y = VA, col = sector)) + 
  facet_grid(.~sector) + 
  geom_point()

enter image description here

Upvotes: 1

Views: 126

Answers (2)

Z.Lin
Z.Lin

Reputation: 29085

If all you want is the mean & standard error bar associated with each "estado"-"sector" combination, you can leave ggplot to do all the work, by replacing the geom_point() line with stat_summary():

ggplot(data,
       aes(x = estado, y = VA, col = sector)) +
  facet_grid(. ~ sector) + 
  stat_summary(fun.data = mean_se)

plot

See ?mean_se from the ggplot2 package for more details on the function. The default parameter option gives you the mean as well as the range for 1 standard error above & below the mean.

If you want to show the original points, just add back the geom_point() line. (Though I think the plot would be rather cluttered for the reader, in that case...)

Upvotes: 2

Sean Lin
Sean Lin

Reputation: 815

Maybe you could try something like below?

set.seed(1)
library(dplyr)
dat = data.frame(estado = factor(rep(1:32, 2)), 
                 sector = rep(letters[1:12], 32), 
                 VA = rnorm(384))
se = function(x) {
    sd(x)/sqrt(length(x))
}
dat_sum = dat %>% group_by(estado, sector) %>%
    summarise(mu = mean(VA), se = se(VA))
dat_plot = full_join(dat, dat_sum)
ggplot(dat_plot, aes(estado, y = VA, color = sector)) + 
    geom_jitter() + 
    geom_errorbar(aes(estado, y = mu, color = sector, 
                      ymin = mu - se, ymax = mu + se)) + 
    facet_grid(.~sector)

enter image description here

Upvotes: 1

Related Questions