Prev: Advanced logical operators?
Next: How to fill in a region with different patterns?
From: Shamim on 21 Feb 2010 14:27
I have trained my data using Steve Gunn's toolbox and now need to generate its ROC curve.
I know one way to generate ROC curves is by changing the bias term in the following code from the Gunn's tolbox (after training the data, I keep alpha0 and ker and vary the bias term):

[nsv0 alpha0 bias0] = svc(trnX,trnY,ker,C);
test_err = svcerror(trnX,trnY,tstX,tstY,ker,alpha0,bias) <--- here the bias will vary

but how to generate the ROC curve, then?! what is the relevant code? What is the range for the bias term??
Which input data is used for generating the curve?!
 | 
Pages: 1
Prev: Advanced logical operators?
Next: How to fill in a region with different patterns?