Fixed variable order in ggpairs

[eibanez]

Hello GGobi group!

My name is Eduardo and I'm one of Dr. Cook's students. Last semester she introduced me to GGally and, as I was using it, I found an error in the way the variables were displayed when using "ggpairs". I finally went through the code and I think I have fixed the issue. I modified the functions "ggpairs" and "plot_types". I believe that these changes don't raise other issues in the code.

You can use the following example to test the changes:

library(GGally)
a <- 1:4
b <- c(10,15,11,14)
c <- c(1000,999,996,990)
data <- cbind(a, b, c)
ggpairs(data)

The current version of "ggpairs" produces the following graphical output. We can check that the axes for each one of the scatterplots are not correct:
http://img863.imageshack.us/img863/4702/ggallyold.jpg

The changes I made in the code fixes the problem:
http://img827.imageshack.us/img827/9139/ggallynew.jpg

You can also verify the change by comparing the output to that of:
plot(data)

A byproduct of this correction is that the ticks in the horizontal axes for the first column are not aligned, but that is a separate problem.

Please let me know your comments.

Eduardo

[hadley]

It's a little clearer to use rev here

[eibanez]

Done!

[hadley]

Looks great otherwise!

[schloerke]

I took a look.

I ran the code you suggested on the latest version:
a <- 1:4
b <- c(10,15,11,14)
c <- c(1000,999,996,990)
data <- cbind(a, b, c)
ggpairs(data)

I had the same output that you want to show in the pull request.

If you took the class last fall (before Dec. 15th, 2010), this makes perfect sense. The error was present then, but it has been corrected since then. I switched the x and y variables around.

I guess I should use more obvious datasets than iris.

THANK YOU for making this aware to me though. I have never done a merge before, but you have set this up very nicely.

Best,
Barret

[eibanez]

I just downloaded the last version (0.2.3) and the output is indeed the intended. I was making my previous changes by executing the code from the old version.

I see what you did in that commit. The reason why you had to switch x and y is that plot_types is not listing the plots correctly. ggpairs expects left-to-right and plot_types cycles top-to-bottom first.

I reverted your change and added it to the pull request. The output is still correct and now x and y are consistent when calling addAndOverwriteAes().

Feel free to either accept or ignore these change. The end result is the same.

[eibanez]

All files are up to date. Please considering merging this.

Thanks!

[jcrowley11]

Sorry it took me so long to get around to this Eduardo. I'm pretty new to GitHub and wasn't really sure what was going on with this "forking" and "pull request," and it took me too long to sit down and read through the guides. Thanks for the contribution!

Jason

[eibanez]

No problem. Learning to use it myselft too. Thanks!