I am using randomForest to classify (relapse vs non-relapse) patients. I
have built a forest using a training data and now want to predict classes in
a test dataset. Instead of using the resulting randomForest object. I was
wondering if there is a way to use the MDSplot. From looking at the MDS plot
it seems like I could draw some lines through the plot to define 'high
risk', 'intermediate' and 'low risk' groups with higher accuracy than what
can be achieved with the forest itself. I have two questions:
1) Is this approach advisable?
2) How can I plot each unknown patient in an MDSplot along with the training
data to determine which class it should belong to? I can feed the training
data along with the unknown patient into predict.randomForest and produce
the dissimilarity matrix. But, MDSplot takes an object of class RandomForest
and this is not what predict.randomForest outputs.
--
Obi L. Griffith, PhD
Post-doctoral fellow
Lawrence Berkeley Laboratory
Life Sciences Division, Cancer & DNA Damage Responses
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
