function errorCount=main()
errorCount=0;%Number of errors made by client program, these are basic sanity checks and no guarantee that code is correct

%Test partitioning
x=[1 0; 2 0; 3 0; 4 0; 5 1; 6 1; 7 1; 8 1];
p=partition(x,[2 2 2 2]);
[r c]=size(p);
errorCount=errorCount+(c~=4);%Should receive 4 partitions
getFractions(x,[0 1]);
f=getFractions(p{1},[0 1]);
errorCount=errorCount+(f(1)~=0.5);%partition should be stratified

%Test cross validation
N=1000; %1000 data points
d=7; % 6 input dimensions, one class value
a=rand(N,d);
a(:,d)=round(a(:,d)); %0 or 1 class value
crossValidationLosses=crossValidate(a,[@buildKNNModel @buildKNNModel @buildKNNModel @buildMyModel],[1 2 3 4],10);
[r numCrossValidationScores]=size(crossValidationLosses);
errorCount=errorCount+(numCrossValidationScores~=4);% Should receive one score per algorithm

function [data, functionHandle]=buildMyModel(trainSet,hyperparameters)
data=3;
functionHandle=@evalMyModel;

function classifications=evalMyModel(data,trainPts)
[N inputDim]=size(trainPts);
classifications=zeros(1,N);

function fractions=getFractions(fullDataset,classes)
[rows numClasses]=size(classes);
for index=1:numClasses
    fractions(index)=getFraction(fullDataset,classes(index));
end

function fraction=getFraction(fullDataset,class)
[N d]=size(fullDataset);
count=0;
for index=1:N
    if (fullDataset(index,d)==class)
        count=count+1;
    end
end
fraction=count/N;