%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Matlab script to perform cross validation on numeric data from agricult.
% purpose: data mining with ANNs/MLPs 
% see http://blog.georgruss.de/?p=54
% -----
% script to generate plots and determine optimal network
% for sports science data
% -----
% Georg Ru\ss
% russ@iws.cs.uni-magdeburg.de
% 2007-12-04
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%% Preparation steps, workspace

% clean workspace
clear all;

% set clock
clock_start = clock;
% seed random for reproducible results
rand('seed',1);

%pause;

% change paths to wherever your data is located and readable to matlab
% uses script readColData from
% http://web.cecs.pdx.edu/~gerry/MATLAB/plotting/loadingPlotData.html#colHeadings
[label,id_column,data]=readColData('BuschschulteNN-100R.csv',26,10,1);

%% data specific stuff
% generate data sets for the nnet to play with

set_1 = data;

%% store dimensions of data for possible later loops
%[number_of_examples_cv,number_of_attributes_kv] = size(crossv);

%% Prepare the data for the neural net
% put input attributes and target values into suitable matrices
% requires selecting and transposing the raw input data

size_set_1 = size(set_1);
%size_set_2 = size(set_2);
%size_set_3 = size(set_3);

%splitCV

set_1_examples = transpose(set_1(:,1:size_set_1(1,2)-1))
set_1_targets = transpose(set_1(:,size_set_1(1,2)))

%set_2_examples = transpose(set_2(:,1:size_set_2(1,2)-1));
%set_2_targets = transpose(set_2(:,size_set_2(1,2)));
%set_3_examples = transpose(set_3(:,1:size_set_3(1,2)-1));
%set_3_targets = transpose(set_3(:,size_set_3(1,2)));
 
% data division stuff
%[set_1_train,set_1_val,set_1_test,set_1_trainInd,set_1_valInd,set_1_testInd] = dividerand(set_1_examples)
%pause

% max size of first hidden layer
max_size_first_layer = 16;
% max size of second hidden layer
max_size_second_layer = 16;

net_1_perf_collection_twod = zeros(max_size_first_layer,max_size_second_layer);
net_1_perf_collection_oned = zeros(max_size_first_layer,1);
%net_2_perf_collection = zeros(max_size_first_layer,max_size_second_layer);
%net_3_perf_collection = zeros(max_size_first_layer,max_size_second_layer);

%create network
for i = 2:max_size_first_layer;
    for j = 2:max_size_second_layer;
        %j=1
        % %%%%%%%%%%%%%
        % network for first data set 
        net_1 = newff(set_1_examples,set_1_targets,[i j]);
        net_1.trainparam.min_grad = 0.01;
        net_1.trainParam.epochs = 500;
        net_1.trainParam.lr = 0.5;
        net_1.trainParam.show = NaN;
        net_1.divideFcn = 'dividerand';
        net_1.divideParam.trainRatio = 38/40;
        net_1.divideParam.testRatio = 1/40;
        net_1.divideParam.valRatio = 1/40;
        net_1.inputs{1}.processFcns = {'mapstd'};
        net_1.outputs{2}.processFcns = {'mapstd'};
        
        % training step
        [net_1_trained,net_1_tr] = train(net_1,set_1_examples,set_1_targets);
        %k=2
        
        % determine mse on test data (from network training)
        net_1_perf = net_1_tr.tperf(1,size(net_1_tr.tperf,2))
        net_1_perf_collection_oned(i,j) = net_1_perf;
        net_1_perf_collection_twod(i,j) = net_1_perf;
        
        % %%%%%%%%%%%%%
        % network for second data set
%         net_2 = newff(set_2_examples,set_2_targets,[i j]);
%         net_2.trainparam.min_grad = 0.001;
%         net_2.trainParam.epochs = 100000;
%         net_2.trainParam.lr = 0.5;
%         net_2.trainParam.show = NaN;
%         % training step
%         [net_2_trained,net_2_tr] = train(net_2,set_2_examples,set_2_targets);
%         % determine mse on test data (from network training)
%         net_2_perf = net_2_tr.tperf(1,size(net_2_tr.tperf,2));
%         net_2_perf_collection(i,j) = net_2_perf;
% 
%         % %%%%%%%%%%%%%
%         % network for third data set
%         net_3 = newff(set_3_examples,set_3_targets,[i j]);
%         net_3.trainparam.min_grad = 0.001;
%         net_3.trainParam.epochs = 100000;
%         net_3.trainParam.lr = 0.5;
%         net_3.trainParam.show = NaN;
%         % network training
%         [net_3_trained,net_3_tr] = train(net_3,set_3_examples,set_3_targets);
%         % determine mse on test data (from network training)
%         net_3_perf = net_3_tr.tperf(1,size(net_3_tr.tperf,2));
%         net_3_perf_collection(i,j) = net_3_perf;
%         
%         
        i
        j
   end
end
clf;

%plot(net_1_perf_collection_oned)
mesh(net_1_perf_collection_twod);
hold on;
%surf(net_2_perf_collection);
%mesh(net_3_perf_collection);

%diff_net_1_net_2 = net_1_perf_collection - net_2_perf_collection;
%diff_net_2_net_3 = net_2_perf_collection - net_3_perf_collection;
%diff_net_1_net_3 = net_1_perf_collection - net_3_perf_collection;
 
% set some training parameters for network training, can be adjusted easily
%net.trainparam.min_grad = 0.000005;
%net.trainParam.epochs = 100000;
%net.trainParam.lr = 0.2;
%net.trainParam.show = 100;

% Start training with the data


duration = etime(clock, clock_start)