hullviz.m

% clear all
% close all
% clc;
% load interdata
% idx = idxfull;
% coord(:,1) = coordfull(:,2);
% coord(:,2) = coordfull(:,1);

function hullviz(idx, coord, clusters)

numclusters = size(clusters,1);

%Initialize now so you can call it in a conditional statement later
vecmat = [];


%%%First find the convexhull so you can draw it

for k = 1:numclusters
    %Clear to avoid mismatch later
    points = [];
    
    index = find(idx == k);
    
    points(:,1) = coord(index,1);
    points(:,2) = coord(index,2);
    
    %%TODO PROBLEMS IF YOU DON'T HAVE THREE UNIQUE <X,Y> IN POINTS, CAN'T
    %%MAKE A CONVEX HULL
    
    m = size(points,1);
    dontcheck = zeros(size(points,1),1);
    unique = m;
    for i = 1:m       
        if dontcheck(i) == 1
            continue
        end
        
        [a, ~] = find(points == points(i,1));
        [b, ~] = find(points == points(i,2));
        
        c = intersect(a,b);
        dontcheck(c) = 1;
        
        unique = unique - (length(c) - 1);
    end
    
    if unique < 3
        vec(size(vecmat,1),2) = NaN;
        vecmat = [vecmat vec];
        continue
    end
    
    
    %%Have to do add a bunch of NaN's because you can't append a column of
    %%different size.
    hull = convhull(points(:,1), points(:,2));
    
    
    %Have to reinitialize because of size problems
    vec = [];
    vec(:,1) = points(hull,1);
    vec(:,2) = points(hull,2);
    
    %%Now aggregate
    
    if k ~= 1 && size(vecmat,1) < size(vec,1)
        vecmat((size(vecmat,1) + 1):size(vec,1), :) = NaN;
    end
    if k ~= 1 && size(vecmat,1) > size(vec,1)
        vec((size(vec,1) + 1):size(vecmat,1), :) = NaN;
    end
    
    vecmat = [vecmat vec];
    
end


%%%PLOT


bg = imread('purduemap.png');

% axes for image
min_y = 40.41678;
min_x = -86.930501;
max_y = 40.438766;
max_x = -86.905546;



% disp image
figure;
imagesc([min_x max_x], [min_y max_y], flipdim(bg,1));
hold on

% Plot the GPS locations and clusters

plot(coord(:,1),coord(:,2),'.')
hold on
%%% TODO label the clusters
plot(clusters(:,1),clusters(:,2),'r*')
title(['GPS location data and ' , num2str(numclusters) , ' clusters'])
xlabel('lat')
ylabel('long')
hold on

% Plot the outside of each cluster

k = 1;
index = ~isnan(vecmat(:,k:k+1));
vectemp = vecmat(1:sum(sum(index,1))/2,k:k+1);

plot(vectemp(:,k),vectemp(:, k+1),'g','linewidth',2)
hold on

% Already did the first cluster
for k = 2:numclusters
    
    index = ~isnan(vecmat(:,2*k - 1:2*k));
    vectemp = vecmat(1:sum(sum(index,1))/2,2*k - 1:2*k);
    
    plot(vectemp(:,1),vectemp(:,2),'g','linewidth',2)
    hold on
end

% flip
set(gca,'ydir','normal');

%TODO Plot transition probabilities, interaction probabilities, etc

end