0

I am creating bar plots in qplot, (ggplot2 function in R) and have two problems.

  1. When I rotate the x-axis scale values by 90 degrees they no longer centre under the columns and tick marks. This doesn't seem like much of an issue with just a few columns, but I'm creating multiple plots with 12 columns using facet-wrap and then the offset becomes very noticeable (the scale values then sit between the columns).

  2. The x-axis values are dates as factors. How can I order the columns by date rather than by alphabetical order. If I convert the dates to date format they will order correctly but will space out on a continuous date scale which I dont want. This could be a problem with any type of x-axis data of course.

Here's a dummy dataset and R code to illustrate the problems.

date <- c("Oct 2006","Feb 2007","Nov 2007","Feb 2008","Oct 2008","Feb 2009","Nov 2009")
count <- c(23,54,12,78,23,34,12)
df <- data.frame(date,count)

qplot(x=date, y=count, data=df, geom="bar", stat="identity") + theme (axis.text.x = element_text(angle = 90,))
fredtantini
  • 15,966
  • 8
  • 49
  • 55
Adrian Pinder
  • 3
  • 1

1 Answers1

1

  1. To make things easy, convert your date strings to a proper date format and then you don't have to worry about ordering them yoruself

  2. by default, the vertical alignment is 1 which puts the tick labels to the right of ticks, 0 to the left, and .5 for centered.

I would also suggest switching to ggplot instead of qplot

date <- c("Oct 2006","Feb 2007","Nov 2007","Feb 2008","Oct 2008","Feb 2009","Nov 2009")
df <- data.frame(
  count = c(23,54,12,78,23,34,12),
  date = factor(d <- as.Date(paste0('01 ', date), '%d %b %Y'),
                labels = format(d, '%b %Y'))
  )


library(ggplot2)
qplot(x=date, y=count, data=df, geom="bar", stat="identity") + 
  theme (axis.text.x = element_text(angle = 90, vjust = .5))

enter image description here

rawr
  • 20,481
  • 4
  • 44
  • 78
  • Thats perfect, thanks very much rawr. I had tried using hjust but it didnt seem to change the position of the scale values, but I guess because the scale values are rotated vjust would apply. I didnt realise that I could convert the date text to as.date then use 'factor'. I tried having the dates in date format but then it spread them out as thin columns across a 3 year continuous x axis. Many thanks. – Adrian Pinder Dec 19 '14 at 00:15