[R] ggplot2: using more than 6 symbols

[Dave Murray-Rust]

On 17 Nov 2008, at 13:00, hadley wickham wrote:

> Hi Dave,
>
> On Mon, Nov 17, 2008 at 6:35 AM, Dave Murray-Rust
> <d.murray-rust at ed.ac.uk> wrote:
>> I'm trying to plot multiple lines using different colours/symbols to
>> distinguish them. If I try to plot more than 6 lines, I get an error:
>>
>>> ggplot( dat, aes(x=time,y=value,group=variable,shape=variable) ) +
>>> geom_line()
>> Error: scale_shape can deal with a maximum of 6 discrete values,  
>> but you
>> have 9.  See ?scale_manual for a possible alternative
>>
>> I find this slightly confusing, as there are ~25 symbols listed for  
>> use.
>> (I'm using ggplot2 0.7, on OSX, R 2.7.2)
>
> But distinguishing between more than 6 of is very difficult!

Absolutely! But it's easier than between different colours of line,  
especially given low-quality printer output...

I guess the "correct" answer is that if you're trying to plot more  
than 6 lines on the graph, you need to rethink the graph.

>
>
>> Ideally, I'd like to use a combination of colours and symbols (e.g. 5
>> colours and 4 symbols -> 20 unique possibilities) or even colours,  
>> symbols
>> and linetypes - is there a way to do this in ggplot2?
>
> Unfortunately not - this type of combined scale is on my to do list,
> but I haven't got around to it yet.  Here's a work around:
>
> df <- data.frame(a = 1:20, b = 1:20, c = letters[1:20])
> qplot(a, b, data = df, colour = c)
>
> df$d <- factor(rep(1:5, each = 4))
> df$e <- factor(rep(1:4, by = 5))
>
> qplot(a, b, data = df, colour = d, shape = e)


Ah, thanks for that; I'd been trying something similar, with slightly  
limited success.

>
>
> Generally, I'd wouldn't advise splitting up a single variable across
> two aesthetics like this: perceptually I don't think its a great idea.

I agree, that is quite confusing - a combined scale would definitely  
be more clear.

Thanks for the info!

Cheers,
dave

-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.