Talk:DataONE:Notebook/Data Citation and Sharing Policy/2010/07/22

re: plotting

Heather A Piwowar 12:56, 22 July 2010 (EDT): Nice work, Nic. A few tips to make these plots easier to read for others, and for embedding in papers, later.
add a title. Right after doing plot(response), do

title("This is a title")

and you should see it appear at the top of the plot

rename your variable. I wasn't very creative in calling the new variable simple.var, so you can do way better :) Use a different variable name for the shared variable ("any.sharing.policy" or somehting) than for the cited instructions variable ( "any.cited.policy" or "has.policy.on.cite.format" or something). That will make the variable at the bottom of your plots (and your code) easier to decypher
save as png and embed. Do you know how to save a figure in R? see

?png

for examples. Then you can upload to OWW, and I think make it show up in the body of your pages using

[[File:Example.png]]

Heather A Piwowar 13:25, 22 July 2010 (EDT): If you want to peak ahead to multivariate logistic regression, here are a few resources to read and work through:
http://www.ats.ucla.edu/stat/R/dae/logit.htm is a very practical example and I imagine that your analysis will progress pretty much as they lay it out. you'd substitute the

 admit~gre+gpa+topnotch

part with

 any.sharing.policy~impact.factor + is.biology

or whatever :) BTW I expect that a lot (most?) of that webpage will be confusing at this point. No fear. Get out of it what you can.

http://www.ats.ucla.edu/stat/stata/faq/oratio.htm for more of a gut feel on what the results mean. Note it gives an example in the stats program Stata not R, but similar.
I'll try to pull up some more intro resources, and/or google "multivariate logistic regression" to find one that speaks to you. Also we'll chat through what it means. Just wanted to give you some links to try to read if you want some exposure....

Heather A Piwowar 13:51, 23 July 2010 (EDT): Nic, I love seeing those embedded plots on your notebook pages. Nice. Want to add a blurb on the newbies page saying how to do it?

Heather A Piwowar 13:51, 23 July 2010 (EDT): when including a link to git, do it like this: http://gist.github.com/487090 where it has an http:// at the front (rather than a git:// ) and no .git at the end. That makes it easily clickable by others
Heather A Piwowar 13:51, 23 July 2010 (EDT): that said, can we embed the gist? maybe we can. let's see:

sweet! Can you add instructions on how to do this to the newbies page as well, please?

Heather A Piwowar 13:58, 23 July 2010 (EDT): looking at your graphs makes me want to have a link to the R code gist that created the graphs AND a link to the spreadsheet/table you used, right there before the graphs. Would make it really clear and very easy for quick skimmer to see and dig into what you are doing. What do you think, doable? You are our current R-in-OWW trailblazer....

Heather A Piwowar 13:58, 23 July 2010 (EDT): would be great if your gist could read data straight from google spreadsheets, rather (or in addition, if sometimes you need to be able to look at it offline) than from your desktop. That would allow it to be much less ambiguous about what dataset you were running, and easier for others to quickly get to the same spot you are at. Let's work on it today to see if we can do it. To facilitate that, can you point me to the url of the data this represents, and make sure that it is as public/open/published as possible?