Togaware DATA MINING
Desktop Survival Guide
by Graham Williams
Google

Further Tuning Models

One of the goals of Rattle is to keep things simple for the user. Consequently, not all options available for many of the functions provided by R are exposed through the Rattle user interface. This is not meant to be a limitation though, and Rattle is quite at ease working with modifications you make to the crs data structure within the R Console, at least to quite some extent.

Suppose for example that you wish to build an ada model using the x and y arguments rather than the formula argument. First, within Rattle, build the normal ada model and go to the Log tab to highlight and copy the command used:

crs$ada <- ada(Adjusted ~ .,
               data=crs$dataset[crs$sample,c(2:4,6:10,13)], 
               control=rpart.control(maxdepth=30, cp=0.010000, 
                                     minsplit=20, xval=10), 
               iter=50)

Now past that into the R Console, but modify it approriately:

crs$ada <- ada(crs$dataset[crs$sample,c(2:4,6:10)], 
               crs$dataset[crs$sample,c(13)], 
               control=rpart.control(maxdepth=30, cp=0.010000, 
                                     minsplit=20, xval=10),
               iter=50)

You can now go back to the Rattle window's Evaluate tab and evaluate the performance of this new model. Indeed, you can, if you choose, save the models to different variables in the R Console, and selectively copy them into crs$ada and then evaluate then with Rattle. Of course, the alternative is to copy the R commands for the evaluation from the Log tab of Rattle and paste them into the R console and perform the evaluation prgrammatically.



Subsections
Copyright © 2004-2010 Togaware Pty Ltd
Support further development through the purchase of the PDF version of the book.
The PDF version is a formatted comprehensive draft book (with over 800 pages).
Brought to you by Togaware. This page generated: Sunday, 22 August 2010