www.gusucode.com > classification_matlab_toolbox分类方法工具箱源码程序 > code/Classification_toolbox/Other/ROCC.m
function [false_alarm, hit] = ROCC(features, targets) %Generate a receiver operating characteristic curve (ROCC) for 1-D data %Inputs: % features - The data from which to estimate % targets - The class for each of the features %Outputs: % false_alarm, hit - The x and y axes for the ROCC. % If the function is called with not output, the ROCC is plotted Nbins = max(3,floor(length(features).^(1/3))); indice0 = find(targets ~= 1); indice1 = find(targets == 1); range = [min(features), max(features)]; p0 = high_histogram(features(:,indice0),Nbins,range); p1 = high_histogram(features(:,indice1),Nbins,range); p0 = p0 ./ sum(p0); p1 = p1 ./ sum(p1); false_alarm = 1-cumsum(p0); hit = 1-cumsum(p1); if (nargout == 0), figure plot(false_alarm, hit) end