% Input data...
Xtrain;  Ytrain;
Xtest;   Ytest; 

% Compute sample covariance matrix
S = cov(Xtrain');
% Note, matlab assumes each column is a variable, thus the transpose

% Compute eigenvectors
[U,D] = eig(S);
U = fliplr(U);  % eig orders eigenvectors from least to most important

% Set level of approximation
k = 5; 

Xtrain_reduced = U(:,1:k)' * Xtrain;
Xtest_reduced  = U(:,1:k)' * Xtest;

[ccMean,ccVar] = Trainer(Xtrain_reduced,Ytrain);

Acc = Tester(ccMean,ccVar,Xtest_reduced,Ytest);

%%%  Of course... you'll want for-loop for various k...