s-news
[Top] [All Lists]

svm() output interpretation

To: S-News <s-news@wubios.wustl.edu>
Subject: svm() output interpretation
From: Kim Elmore <Kim.Elmore@noaa.gov>
Date: Thu, 13 Dec 2007 15:32:36 -0600
User-agent: Thunderbird 2.0.0.9 (Windows/20071031)
I'm working with svm() in the svm library. I'm using it to generate a classifier for four classes. Problem is, I'm not sure how to interpret the output when I ask for cross-validation. The documentation refers to "accuracy," but I'm used to "accuracy" for a given category as something like number.correct/total.placed.in.class. For example, the total "accuracy" of a model might be total.correctly.classified/total.number.of.cases. There are lots of variations, but that's the idea. In all cases, these numbers will be at most 1, and almost always less. However, I see numbers that look more like 100*accuracy. Is this a correct interpretation? What am I missing?

--

Kim Elmore, Ph.D. (PP SEL/MEL/Gllider, N5OP, 2nd Class Radiotelegraph, GROL)
"All weather is divided into three parts: Yes, No and Maybe. The greatest of these 
is maybe." -- The original Latin appears to be garbled.


<Prev in Thread] Current Thread [Next in Thread>
  • svm() output interpretation, Kim Elmore <=