Shade several areas of a time series plot: error using geom_rect

Question

I am trying to shade several sections of a time series plot using “geom_rect”. However, I have the error message “Error in as.Date.default(date) : do not know how to convert 'date' to class “Date”. Why ? and how to fix this ?

Here is a reproducible example:

  library(ggplot2)
  dat <- data.frame(date=seq.Date(as.Date('2005-01-01'), as.Date('2016-12-31'),length.out=100), y=runif(100)) 
  print(dat)
  block <- data.frame(level = c('A','B', "C"), 
                       ymn = -Inf, 
                       ymx = Inf,
                       xmn = c(as.Date('2006-01-01'), as.Date('2010-01-01'), as.Date('2014-01-01')),
                       xmx = c(as.Date('2006-12-31'), as.Date('2010-12-31'), as.Date('2014-12-31')))
  print(block)

  ggplot(data=dat, aes(x=as.Date(date), y=y)) + 
    geom_line(size = 1) +
    geom_rect(data=block,aes(xmin=as.Date(xmn),xmax=as.Date(xmx),ymin=ymn,ymax=ymx, fill=level), alpha = 0.3)

You might need an `inherit.aes = FALSE` in `geom_rect`. It might be seeing the global `x` you defined. — aosmith, Jun 07 '18 at 20:56
The column `date` already is a Date object, as are the date columns in `block`. Take out the `as.Date` calls in all your `aes` and this works — camille, Jun 07 '18 at 21:05
Great ! it works by adding `inherit.aes = FALSE` in `geom_rect` ! Thanks very much for your help ! — Nell, Jun 07 '18 at 22:20

score 0 · Accepted Answer · answered Jun 07 '18 at 20:57

0

This will work:

ggplot(data=dat, aes(x=as.Date(dat$date), y=dat$y))

or

ggplot(data=dat, aes(x=dat$date, y=dat$y))

The next option will work as well:

ggplot(data=dat, aes(x=as.Date(dat$date), y=dat$y)) + 
  geom_line(size = 1)

but your real problem is that the block object is not the right shape to match data the way you're using it, but that's an unrelated problem.

answered Jun 07 '18 at 20:57

Hack-R

22,422
14
75
131

3

Referencing the name of the data frame inside `aes`, such as `y = dat$y`, can be really prone to bugs – camille Jun 07 '18 at 21:03
@camille Could you be specific? I've never experienced any bug – Hack-R Jun 07 '18 at 21:16
There are a lot of SO posts where something weird happens in a ggplot, and it turns out the issue is a scoping problem introduced by using the dataframe name inside `aes`. ggplot expects to get bare column names, so you're better off feeding it what it wants. Here's one post illustrating this: https://stackoverflow.com/questions/22244500/why-ggplot2-pie-chart-facet-confuses-the-facet-labelling/22245032#22245032 – camille Jun 07 '18 at 21:32

Shade several areas of a time series plot: error using geom_rect

1 Answers1