[R] Negative/ positive rolling correlation

David Winsemius dwinsemius at comcast.net
Thu Mar 13 14:57:01 CET 2008


Rthoughts <seb.brockel at northampton.ac.uk> wrote in
news:16013887.post at talk.nabble.com: 

 
> There is one question I have though. I can see that worksheets and
> history can be saved, however I am beat as to exactly what they
> save? Do they save the command lines and associated data sheets and
> graphs? 

There is no "datasheet" but the R objects defined and saved in the last 
session will automatically be loaded into the workspace from the .Rdata 
file when a new session begins. It is also possible to save named .Rdata 
files and reload them under user control. If you look at an .Rdata file 
with a text editor, it is pretty much meaningless. On the other hand, if 
you look at .Rhistory, you will see a text file that is simply a 
sequential listing of the commands of earlier saved sessions.
> 
> Oh, I'm using Rcmdr. In my case it is an excellent software that
> cuts hours off my progamming work freeing up time for other parts of
> the project and breaks, lol!
> 
> The other, I want to do a correlation to highlight matching patterns
> between two columns of radon levels (96 lines each) divided by the
> hour they were recorded in the two machines. A scattergraph is
> similar but I want to produce visualtisations for negative/ positive
> correlation against time. Can anyone suggest anything? Thanks.

You need to investigate time series analyses methods. I am assuming that 
the radon levels are ordered and associated with times, rather than 
mathematically divided by them. There are extensive capabilities for 
analyzing and displaying such information in R.

See for instance:
<http://cran.r-project.org/doc/contrib/Ricci-refcard-ts.pdf>

-- 
David Winsemius



More information about the R-help mailing list