Prev: Selecting a Region in an Image based on Color
Next: Passing objects with fminsearch
From: MatlabBo St on 9 Dec 2009 15:55
Hi there,

could anybody help me understand the right approach of using the ROC curve?

I have labels '0' and '1', and truePositivies and falsePositivies. I would like to use this:

[Xplot,Yplot] = perfcurve(truePositivies,falsePositivies,'1');

but I dont know what kind of data do I need to feed the function.

Simply,

truePositivies = (Y_predicted == Y);
falsePositivies = (Y_predicted ~= Y);

doesnt help.

Thank you very much in advance !
From: MatlabBo St on 10 Dec 2009 14:02
Ok, a revised question:

lY = length(Y);
for i = 1:10
c = confusionmat(Y_predicted(i),Y);
TPR(i) = c(1,1)/lY;
FPR(i) = c(1,2)/lY;
end

I still cannot draw the ROC curve with: perfcurve(TPR,FPR,'1');

What am I missing / not understanding ?
From: MatlabBo St on 10 Dec 2009 16:35
Maybe I should a more general question:

How do we use the input parameters of the perfcurve funciton ?

labels - labels can be a numeric vector, logical vector, character matrix, cell array of strings or categorical vector.
scores - scores is a numeric vector of scores returned by a classifier for some data. This vector must have as many elements as labels does.
posclass - posclass is the positive class label (scalar), either numeric (for numeric labels) or char. The specified positive class must be in the array of input labels.


Suppose, I have a classfier which has two classes: 0 and 1 ... How would the input params look like so that I can draw the ROC curve afterwards ?

Best regads,
From: MatlabBo St on 10 Dec 2009 18:09
P.S. Another good example would be a perfcurve used to draw this:

http://gim.unmc.edu/dxtests/ROC2.htm
 | 
Pages: 1
Prev: Selecting a Region in an Image based on Color
Next: Passing objects with fminsearch