prtScoreRocBayesianBootstrapNfa   Generate a reciever operator characteristic curve with Bayesian Boostrapping
 
    prtScoreRocBayesianBootstrapNfa(DECSTATS,LABELS) plots the receiver
    operator characteristic curve for the decision statistics DECSTATS and
    the corresponding labels LABELS. DECSTATS must be a Nx1 vector of
    decision statistics. LABELS must be a Nx1 vector of binary class
    labels. This behavior is the same as prtScoreRocBayesianBootstrap. The
    only difference is when output arguments are requested (see below).
 
    prtScoreRocBayesianBootstrapNfa performs Bayesian boot strap sampling
    of an ROC curve and generates the 100*(1-alpha) percent percentile
    uniform credible band. The default alpha is .05, corresponding to a 95%
    credible band. This is done following the methodology in:
 
    Non-parametric estimation of ROC curve J. Gu, S. Ghosal, and A. Roy
    Statistics in Medicine, Vol. 27, 5407—5420, 2008.
    http://www4.stat.ncsu.edu/~ghosal/papers/ROCBB.pdf
 
    prtScoreRocBayesianBootstrapNfa(DECSTATS,LABELS,
    NBOOTSAMP) Specifies the nummber of boostrap samples NBOOTSAMP. The
    default value is 1000.
 
    prtScoreRocBayesianBootstrapNfa(DECSTATS,LABELS, [],
    NPFSAMP) Specfies the number of samples of probability of false alarm
    at with which to sample the ROC curve. The default is 500.
 
    prtScoreRocBayesianBootstrapNfa(DECSTATS,LABELS, [],
    [], ALPHA) Specifies ALPHA, the size of the credible interval
    100*(1-alpha). The default is 0.05, corresponding to a 95% credible
    band
 
    [NFA, PDMEAN,PDCONFREGION, BOOTSTRAPPEDPDS] =
    prtScoreRocBayesianBootstrapNfa(...) outputs NFA, the
    number of false alarms at which the bootstrapped ROC curves are
    evaluated. PDMEAN, the mean of the bootstrapped ROC curves,
    PDCONFREGION, the 100*(1-alpha) percent percentile uniform credible
    band reported as the upper and lower Pd curves. BOOTSTRAPPEDPDS, all
    samples of the bootstrapped ROC curves
 
     Example:     
     TestDataSet = prtDataGenBimodal;       % Create some test and
     TrainingDataSet = prtDataGenBimodal;   % training data
     classifier = prtClassSvm;             % Create a classifier
     classifier = classifier.train(TrainingDataSet);    % Train
     classified = run(classifier, TestDataSet);     
     % Find the number of false alarms at the corresponding PD.
     [nf, pd]= prtScoreRocBayesianBootstrapNfa(classified.getX, TestDataSet.getY, [],[],.2);