% author: M.M.Haji
% email : mehdi.haji@gmail.com
%
% make a decision tree to detect text, based on dct-18 coefficients.

clear

disp('read training data...');
D = textread('textdata.txt');

disp('shuffle training data...');
D = D(randperm(size(D,1)),:);

trset = D(1:1000,:);

[nr nc] = size(trset); % nr == 19

targets = repmat('N',1,nr);
targets(find(trset(:,nc))) = 'Y';

disp('learning the decision tree...');
text_dtree = treefit(trset(:,1:nc-1),targets');

disp('test the decision tree...');
[c,s,n,best] = treetest(text_dtree,'cross',trset(:,1:nc-1),targets');

disp('prune the decision tree...');
text_dtree = treeprune(text_dtree,'level',best);

clear D trset nr nc targets c s n best;
save text_dtree;