analyse_spawns.m

% this .csv was exported from pokeminer
T = readtable('./pokeloc.csv');

spawnpoints = T{:,[6,7]};
spawnpoints = unique(spawnpoints,'rows');

spawnrates = zeros([size(spawnpoints,1),150]);
spawnnumber = zeros([size(spawnpoints,1),1]);

for ispawn = 1:size(spawnpoints,1)
    ispawn
    idx = sum(repmat(spawnpoints(ispawn,:),[size(T,1),1])==T{:,[6,7]},2)==2;
    spawnnumber(ispawn) = sum(idx);
    for jspawn = find(idx)'
        spawnrates(ispawn,T{jspawn,2}) = spawnrates(ispawn,T{jspawn,2}) + 1/spawnnumber(ispawn);
    end
end

%%
  
idx = spawnnumber>40;
pcs = 4;
ncluster = 4;
[coeff1, score1,latent,tsquared,explained,mu1] = pca(spawnrates);
[IDX,C] = kmeans(score1(:,1:pcs),ncluster);
cmap = lines(7);
cmap(4,:) = cmap(1,:);
cmap(1,:) = cmap(5,:);
clear I

for idx = unique(IDX)';
    I(idx,:) = IDX == idx;
    hold on
end

%%
clusters = {'bugs','thrash','parks/nests/rare','river','','','',''};
figure
hold on
for idx = unique(IDX)'
    plot3(score1(I(idx,:),1),score1(I(idx,:),2),score1(I(idx,:),3),'o','Color',cmap(idx,:))
end
xlabel('pc1')
ylabel('pc2')
zlabel('pc3')
box on
legend(clusters)

figure
hold on
for idx = unique(IDX)'
    plot3(score1(I(idx,:),1),score1(I(idx,:),2),score1(I(idx,:),4),'o','Color',cmap(idx,:))
end
legend(clusters)

figure
hold on
for idx = unique(IDX)'
    plot3(score1(I(idx,:),1),score1(I(idx,:),3),score1(I(idx,:),4),'o','Color',cmap(idx,:))
end
legend(clusters)

figure
hold on
for idx = unique(IDX)'
    plot3(score1(I(idx,:),2),score1(I(idx,:),3),score1(I(idx,:),4),'o','Color',cmap(idx,:))
end
legend(clusters)


figure
gscatter(spawnpoints(:,2),spawnpoints(:,1),IDX,cmap)
legend(clusters)

t = C*coeff1(:,[1:4])' + repmat(mu1,size(C,1),1);

%%
% this is the pokemon.json for pokemon names
pokedex = loadjson('./pokemon.json');

fid = fopen('clusteranalysis.txt','w');
for iclust = 1:ncluster
    fprintf(fid,['cluster ' num2str(iclust) ': ' clusters{iclust} ' (' num2str(sum(IDX==iclust)/length(IDX)*100,'%0.1f') '%%)\n']);
    pidx = find(t(iclust,:)>0.01);
    for ipoke = pidx
        fprintf(fid,[pokedex{ipoke}.Name ': ' num2str(t(iclust,ipoke)*100,'%0.1f') '%%\n']);
    end
    fprintf(fid,['other: ' num2str((1-sum(t(iclust,pidx)))*100,'%0.1f') '%%\n\n\n']);
end
fclose(fid)