Statisticus aims to do three things well:
Integrate other workflow tools to statistical analysis: memoization, composition, concurrency, and distributed computing
Remove seams from the R runtime and the R bindings (RSRuby)
Bridge simple Ruby-centric data containers (hashes, 2-dimensional arrays) into R-centric data containers (lists and data frames)
The most basic usage can be:
# create a class, GeometricMean # look for geometric_mean.r somewhere in this gem, in the working directory where the Ruby process is started, or in the ~/.statisticus directory # expect geometric_mean to be a defined function in geometric_mean.r
If that holds true, then you've got the easiest access to the R runtime I can think of.
Since I'm using TeguGears here, I have a few other tricks baked in:
The calls are memoized. So, GeometricMean.call() gets stored in a central repository which can be balanced, pruned, and managed if you'd like.
The calls can be composed, so things like (SquareRoot | GeometricMean ).call() will work.
Other features are on their way, just not all tested, like concurrency and distributed processing.
A more interesting example might be:
class Whatever # Fill in the gaps here. end
Decide if I want to move my standard stats libs over here. Possibly all of Sirb as well?
Put all of these into Panorama
Confirm run vs process (from TeguGears)
If you have R code:
If you want to pass the code along:
Dealing with parameter lists
Running this in parallel, memoization
sudo gem install davidrichards-statisticus
Copyright © 2009 David Richards. See LICENSE for details.