circles and arrows

[Roney Fraga]

Hello all,

I'm making a chart with the following code:

# ------------
# begin
# ------------
library(ggplot2)
library(grid)
mydf <- read.csv(file='http://dpaste.com/24KGSW9.txt', sep=',')
head(mydf)
g <- ggplot(data=mydf, aes( x=t.ano, y=nn.t.grupo, size=t.ttgrupo)) +
geom_point(shape=21) +
scale_size_area(max_size=12) +
scale_x_continuous(limits=c(min(mydf[,'t.ano']), max(mydf[,'t.ano'])), breaks = seq(min(mydf[,'t.ano']), max(mydf[,'t.ano']),by=1)) +
scale_y_continuous(limits=c(min(mydf[,'nn.t.grupo']), max(mydf[,'nn.t.grupo'])), breaks = seq(min(mydf[,'nn.t.grupo']), max(mydf[,'nn.t.grupo']),by=1)) +
geom_segment( data=mydf, aes( x = xbeg, y = ybeg, xend = xend, yend = yend),
             arrow = arrow(length = unit(0.30,"cm"), type='closed'), size=0.5, color='black') +
theme(legend.position='none')
print(g)
# ------------
# end
# ------------

The link to the graph is here.
http://i.imgur.com/o0nyYzA.png

I need to start and stop the arrows on the edges of the circles, given the size of the circles depends on the variable mydf$t.ttgrupo. I would have to know the radius of the circles to discounts them in variables mydf$xbeg and mydf$xend. Any tips on how to do this?

Best
Roney

[Ben Bond-Lamberty]

As always I'd prefer to work with a simpler example first. Is this the
effect you want?

d <- data.frame(x=1:3,z=c(1,18,3))
d$xend <- c(d$x[-1], NA)
d$nextz <- c(d$z[-1], NA)
d$xend <- d$xend - d$nextz/750 # this is a scaling factor, I just
fit by trying different values
p <- ggplot(d, aes(x,x)) + geom_point(shape=21,aes(size=z))
p + geom_segment(aes(xend=xend,yend=xend),arrow = arrow(length =
unit(0.30,"cm"), type='closed'))

The cheat here is that 'scaling factor', which should be computed
programatically from the ggplot size scale...but I just tried various
values to find one that looked good. This may or may not work for you.

Ben

> --
> --
> You received this message because you are subscribed to the ggplot2 mailing
> list.
> Please provide a reproducible example:
> https://github.com/hadley/devtools/wiki/Reproducibility
>
> To post: email ggp...@googlegroups.com
> To unsubscribe: email ggplot2+u...@googlegroups.com
> More options: http://groups.google.com/group/ggplot2
>
> ---
> You received this message because you are subscribed to the Google Groups
> "ggplot2" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to ggplot2+u...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[Roney Fraga]

Ben, his example made ​​the problem clearer. Thank you! 

This solution is good, but I would have to adjust the scale factor for each database that I am using. It would be great to have a general solution. 

Best, 

Roney

[Ben Bond-Lamberty]

I bet you could do something along these lines to extract the size
scale information:

http://stackoverflow.com/questions/7705345/how-can-i-extract-plot-axes-ranges-for-a-ggplot2-object

For example,

gp <- ggplot_build(p)
sizerange <- gp$plot$scales$scales[[1]]$range$range

...and then use this information to adjust the scaling factor in my
code example. (Sorry, busy day, can't work this out myself right now.)

Ben

[Brandon Hurr]

I've done something similar before... 

p<-ggplot(data=mtcars, aes(x=mpg))+geom_histogram()

#dissect the guts for the range (default bins)

p.build<-ggplot_build(p)

p.build$panel$ranges[[1]]$y.range[2]

q<-p+scale_y_continuous(limits = c(0,p.build$panel$ranges[[1]]$y.range[2]*1.2))

[Roney Fraga]

I will work tonight in this suggestion.

Thanks.

[Roney Fraga]

I standardized the group to have 1 as a maximum value: 

    df2$t.ttgrupo.p <- df2$t.ttgrupo/max(df2$t.ttgrupo)

and now all the arrows and circles have the same proportions.

Thanks for the help. 

Roney