function parameters=modelData(data,dim,nc,steps,squeeze,howmuch,dim1,dim2)

%This function simply strips dim eigenvectors off the Data
%Calculates an initial estimate for the nc means of the Gaussian mixtures
%Hands the Data off to the EM algorithm for iterative refinement of the 
%model parameters


maxObs = max(data);
minObs = min(data);
meanObs = mean(data);

%Now we can get any number of mean estimates by interpolating from the
%maximum observation to the minimum observation.
%Hey, its an estimate!


if(squeeze==1)
%Squeezing the max and min observations towards the mean in an attempt to 
%avoid modeling outliers
minObs = minObs + howmuch*(meanObs-minObs);
maxObs = meanObs + (1-howmuch)*(maxObs-meanObs);
end

%The mean is not likely to be the halfway interpolation between min and max,
%but unless the data has a significant skew, its likely to be close.

for mn=1:nc
   means(mn,:)=minObs + ((mn-1)/(nc-1))*(maxObs-minObs);
end


%Make sure the sample mean is one of the estimates
means(ceil(nc/2),:) = meanObs;
end

alphas = ones(1,nc)*1/nc;

parameters = EMND(data,means,alphas,nc,steps,0,1,dim1,dim2)