Adding error bars to a ggplot with reshaped data

Question

I am trying to add error bars to a geom_line() plot which has multiple y values.

To manipulate the ggplot2 y values I had to reshape the data frame in to long format and so the data is structured as shown below.

Here is dput() of my data:

mydata.m <- structure(list(Date = structure(c(16968, 16969, 16970, 16971, 
                                   16972, 16973, 16974, 16975, 16968, 16969, 16970, 16971, 16972, 
                                   16973, 16974, 16975), class = "Date"), error = c(NA, 4e-04, NA, 
                                                                                    0.0085, 0.0106, 0.179, NA, 0.0065, NA, 6e-04, NA, 0.007, NA, 
                                                                                    0.0129, NA, NA), variable = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 
                                                                                                                            1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L), .Label = c("c", "cal5C"
                                                                                                                            ), class = "factor"), value = c(NA, 0.0065, NA, 0.0625, 0.089, 
                                                                                                                                                            0.1825, NA, 0.1299, NA, 0.0046, NA, 0.082, NA, 0.16, NA, NA)), .Names = c("Date", 
                                                                                                                                                                                                                                      "error", "variable", "value"), row.names = c(NA, -16L), class = "data.frame")

The data should look like this:

  head(mydata.m)
        Date  error variable  value
1 2016-06-16     NA        c     NA
2 2016-06-17 0.0004        c 0.0065
3 2016-06-18     NA        c     NA
4 2016-06-19 0.0085        c 0.0625
5 2016-06-20 0.0106        c 0.0890
6 2016-06-21 0.1790        c 0.1825

using ggplot to plot my data:

plot1 <- ggplot(mydata.m[!is.na(mydata.m$value), ], 
            aes(x=Date, y=value, color=variable, group = variable))
plot1 <- plot1 + geom_point(size=8) + geom_line(linetype = 6, lwd =1.5)               

plot1 <- plot1 + scale_color_manual(name="", values = 
                                  c("navyblue","turquoise3"), labels = c("C", "calC")) 

plot1 <- plot1+ theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(),
                  panel.background = element_blank(), axis.line = element_line(colour = "black"))

break.vec <- c(as.Date("2016-06-16"),
           seq(from=as.Date("2016-06-16"), to=as.Date("2016-06-23"), by="day"))

plot1 <- plot1 + scale_x_date(breaks = break.vec, date_labels = "%d-%m",expand = c(0.05,0))

plot1 <- plot1 + theme(text = element_text(size=25), axis.text.x = element_text(size=35), 
                   axis.title.x = element_text(size=45), 
                   axis.title.y = element_text(size=45,margin=margin(t=0,r=20,b=0,l=0)),
                   axis.text.y = element_text(size=35))

plot1 <- plot1 + theme(legend.justification = c(1, 1), legend.position = c(0.05, 1), legend.key = element_rect(fill = "white"))

This makes a fairly appealing graph where i have been able to adjust colours and legend position, which without reshape i was struggling to do. But the problem i now face is how do i add the error to the graphs?

*Note the variables have different numbers of NA values with cal5C having more frequent NAs.

I have tried:

plot1 <- plot1 + geom_errorbar(data =mydata.m[!is.na(mydata.m$error), ], aes(ymin=mydata.m$value - mydata.m$error, ymax=mydata.m$value + mydata.m$error), width=.05)

But I get the following error:

Error: Aesthetics must be either length 1 or the same as the data (32): ymin, ymax, x, y, colour, group

Is there another way to add error bars to both y variables following this format where the values are uneven as are the errors???

Thanks in advance and I hope that makes sense.

Sarah Sarah · Accepted Answer · 2018-01-16T05:22:11

You are referencing the full dataframe in the aes of the geom_errorbar eg mydata.m$erroreven though you've told it to use a cut down dataframe. You should just be referring to the column names

plot1 <- plot1 + 
    geom_errorbar(
        data =mydata.m[!is.na(mydata.m$error), ], 
        aes(ymin=value - error, ymax=value + error), 
        width=.05)

I've also assumed you meant ymax = value + error, not error+ error as you wrote

Please note I haven't checked and run this. In future best to provide your example data with dput(mydata.m) so that its easier for other people to get your data (or an appropriately sized subset) into R to test.

Adding error bars to a ggplot with reshaped data

1 Answers