Run_Pareto.m

function [] = Run_Pareto(dirstring, beh_type, filepath, file_ext, mode, ...
kernel)
    % This function sets up a Pareto Analysis to run for eligible
    % individuals.
    % Parameters:
    % dirstring: string to load from and save to
    % beh type: "AGG","SIB",or "BOTH", the type of behavior
    % filepath: file path leading to the individual model files
    % file_ext: file name of the individual model files
    % mode: 'linear' or 'logistic' activation function
    % kernel: 'yes' or 'no', direct kernel transformation or not
    
    % Set random seed
    rng(0,'twister')
    tic

    % Load all people
    load(strcat(dirstring,'/', beh_type,'people2.mat'));

    for person = 1:length(eligible_people)
        curr_person = eligible_people(person);
        load(strcat(filepath,beh_type,file_ext,string(curr_person),'.mat')) ;
        if WT1{1}.Best == -1
            continue
        end

        bcrs = NaN(1,30); 
        for i=1:30
            bcr = WT1{i}.BCR;
            bcrs(i) = bcr;
        end

        % Sort the BCRs ascending
        [B, I] = sort(bcrs);
        % Find the indices that are nan
        nanB = isnan(B);
        % Get rid of the NaN indices
        I = I(~nanB);
        B = B(~nanB);

        % Find the median of the valid BCRs
        med = median(bcrs,'omitnan');
        med
        % Find the index of the value closest to this median
        [c ind] = min(abs(B-med));
        
        % Find the index in terms of the original indices
        ind = I(ind);
        
        % Get the model at that index
        Model = WT1{ind};
        
        
        % Run the pareto analysis
        
        Model = ParetoAnalysis_Features(Model, Model.Xtrain, ...
            Model.Prototypes, mode, kernel, Model.sigma);
        STDs(person,:) = Model.Var;
        yhats{person} = Model.yhats;
    end
    
    % Save out data
    save(strcat(dirstring, '/', beh_type,'STDs.mat'),'STDs');
    save(strcat(dirstring, '/', beh_type,'yhats.mat'),'yhats');
    
end