Map straight shaded error onto a coord_polar plot

Question

I'd wondering if someone could please offer me a suggestion on how to sort out my shaded error for each line in this radar plot? I've tried several different approaches but am not getting what I want.

library(ggplot2)
library(scales)

# Make some data
Group.no    <- 3
Group.names <- c("1","2","3")
Metric.no   <- 4
Metriclist  <- c("M1", "M4", "M6","M8")

Metric  <- c(rep(c(Metriclist), each = Group.no))
Group   <- c(rep(c(Group.names), times = Metric.no))

Mg    <- c(87.7, 93.8, 72.5, 190.3, 170.9, 138.4, 283.2, 248.7, 196.5, 340.6, 307.9, 240.9)

d <- data.frame(Metric, Group, Mg)
d$lowCI  <- Mg-8
d$highCI <- Mg+8 

# Plot data
Plot <- ggplot(d, aes(x = Metric, y = Mg, group = Group)) +
  geom_polygon(aes(group = Group, colour = Group), fill = NA, size = 1.1) +  
  geom_ribbon(aes(x=Metric,y=Mg,ymin=lowCI,ymax=highCI, group = Group, fill=Group), alpha=.3) +
  coord_polar(start = -((180/Metric.no)*(pi/180)))+
  theme_light()

Plot

Example plot

As you can see, the geom_ribbon isnt plotting the lower and upper CI's in line with geom_polgyon. I am currently at the peak of my knowledge on this. Is anyone able to offer a suitable fix so that the shaded CI's track the polygon please so that both are straight lines?

Thanks in advance for any solutions!

Can you clarify what you would like to be different? Do you want the polygon curved like the ribbons, the ribbons straight like the polygons, or something else? — Miff, Mar 23 '21 at 12:05
@Miff sorry if this wasn't clear. it's for the ribbons to be straight like the polygons are. the straight lines are correct, but the geom_ribbon is curved when I want it straight aswell — Maylor, Mar 23 '21 at 12:15
Possible duplicate: https://stackoverflow.com/questions/42562128/ggplot2-connecting-points-in-polar-coordinates-with-a-straight-line-2 — teunbrand, Mar 23 '21 at 14:29
Also related: https://stackoverflow.com/questions/66196451/draw-straight-line-between-any-two-point-when-using-coord-polar-in-ggplot2-r/66196752 and https://stackoverflow.com/questions/57209060/how-to-draw-a-radar-plot-in-ggplot-using-polar-coordinates/57209765 — teunbrand, Mar 23 '21 at 14:30

score 0 · Accepted Answer · answered Mar 23 '21 at 15:03

With a bit of fiddling with the data, you can just about make it work. Create a separate dataframe that has the confidence intervals (note that we need to repeat the first point to make the intervals work correctly between M8 and M1) with:

i <- c(seq_along(d[[1]]), which(d$Metric=="M1"))

dCI <- rbind(data.frame(d[i,1:2], CI=d$lowCI[i], type="low") ,
             data.frame(d[i,1:2], CI=d$highCI[i], type="high"))

Then we can add the confidence intervals as a polygon with:

Plot <- ggplot(d, aes(x = Metric, y = Mg, group = Group)) +
  geom_polygon(aes(group = Group, colour = Group), fill = NA, size = 1.1) +  
  geom_polygon(aes(x=Metric,y=CI, group = Group, fill=Group), data=dCI, alpha=.3) +
  coord_polar(start = -((180/Metric.no)*(pi/180)))+
  theme_light()

Plot

This produces the output:

Thanks so much @Miff, I had tried adding geom_polygon twice with the fill imputed, but it hadn't crossed my mind to use a different df for this - This works perfectly. Thanks! — Maylor, Mar 23 '21 at 15:18

Map straight shaded error onto a coord_polar plot

1 Answers1