How to obtain AUC for different sample sizes of training data and do comparisons with sensitivity and specificity?

Hello,

I need some urgent advice. WHen I have a SVM classifier implemented, I want to make conclusions from this by varying training and test sets. Is this a valid experiment to show the effect of sample size on sensitivity and specificity and accuracy.

First I tried to do 50% training and testing with cross validation, but due to random selection I iterated it 10 times so I could obtain an average. Then I obtained specificity values that varied so much and the AUC wasn't even correlating with it. How can I obtain a good comparison study.

here is what tried:

 MyData=load ('My_69_Fotos.csv');                             %# load iris dataset

labels=[1;1;1;1;1;1;1;1;1;1;1;1;1;1;1;1;1;... 1;1;1;1;1;0;0;0;1;0;1;0;0;1;0;... 0;0;0;1;0;0;0;0;0;0;1;0;0;... 0;0;0;1;1;1;1;0;0;1;1;0;0;0;0;0;0;... 0;0;0;1;1;0];

amountExudates = MyData(:,1); %ATTRIBUTE1

total_area=MyData(:,2); %ATTRIBUTE2

densityRatio=MyData(:,3); %ATTRIBUTE3

gemiddeld=mean2(densityRatio);

standaard_afwyking=std2(densityRatio);

NormalizedArray =double( (densityRatio -gemiddeld) ./ standaard_afwyking);

GS = NormalizedArray ./ max(NormalizedArray(:))-min(NormalizedArray(:));

TrainVec=[amountExudates ,GS];

[train, test] = crossvalind('holdOut', 68, 0.75); %# get indices

cp = classperf( labels) ; %# init performance tracker

%# train an SVM model over training instances

    svmModel = svmtrain(TrainVec(train,:), labels(train,:), ...

                 'Autoscale',true, 'Showplot',true, 'Method','QP', ...

                 'BoxConstraint',2e-1, 'Kernel_Function','rbf', 'RBF_Sigma',1);

%# test using test instances

pred = svmclassify(svmModel, TrainVec(test,:), 'Showplot',true);

TestingLabel=labels(test);

cp = classperf(cp, pred, test);

x=([GS,test]);

thresholds=3;

alpha=0.05; %significance level

verbose=1;

 ROCout=roc(x,thresholds,alpha,verbose)