An oft heard issue with R is the learning curve. Yet R is a very
powerful language for data mining, if only one persists with it.
Attempting to provide a fast track for anyone to learning R through a
GUI, without limiting the user to just the GUI, I've pulled together
some basic R functionality for "typical" data mining into a GUI
written in R using RGtk2 (need the RGtk2 package from
http://www.ggobi.org/rgtk2/). This differs from other R GUIs in that
it is specifically for data mining type tasks and maps to how we
typically proceed through a project.
The GUI is simple and basic, covering the common tasks of loading a
CSV file, selecting Variables, sampling the data, summarising the
data, clustering, model building (decision tree, regression, random
forrest, boosting, svm), and evaluation (confusion table, ROC, Risk
Chart). Everything is logged to a Log window and direct cut-n-paste
from there to the R Console should work. This provides "tuition" or a
reminder of how to do things.
I've not made it into a formal R package yet but as time permits will
do so. The package has been available and in use in a number of data
mining projects for a couple of months now, and is under continuing
evolution. It works well for what it was designed for (basically
binary classification) but should handle anything thrown at it (at
least gracefully informing you if it can not).
It is freely available (GPL) from
http://www.togaware.com/datamining/rattle.html
Comments, suggestions, bugs, and code are always welome.
Regards,
Graham Williams
