%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Matlab script to perform cross validation on models generated for 
% numeric data from agriculture;
% yield prediction task = multi-dimensional regression
%
% purpose: data mining with SVMs/MLPs/RBFs/RegTree
%
% compares: 
% - support vector regression
% - multi-layer perceptron
% - radial basis function network
% - regression tree
%
% conference: ICDM 2009 http://www.data-mining-forum.de/icdm2009.php
% description: http://blog.georgruss.de/?p=145
%
% requires:
% - statistics toolbox for cvpartition (from Matlab2008a)
% - neural network toolbox
% - SVMTorch (compiled version, of course)
% - data structure used for the results (cv_results)
% - savefig script from matlab central
%   (mathworks.com/matlabcentral/fileexchange/loadFile.do?objectId=10889)
% - a local (user's) .matlab directory 
% - SOM toolbox for data access:
%   (http://www.cis.hut.fi/projects/somtoolbox/)
% - modelcomparison.m script (link must be in blog post or nearby)
% 
%
% new:
% - comparison of SVM result vs. NNet result vs. ...
%   RBF result vs. regtree result
% - included normalization 
% - figures are stored as eps and jpg automagically
% - workspace/model results are stored for later usage
%
% -----
% Georg Ru{\ss}
% russ@iws.cs.uni-magdeburg.de
% 2009-01-26
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%% Preparation steps, workspace

% clean workspace
clear all;

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

% 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('sorted_all',10,10,1);

% data read into struct using SOM toolbox
sD330=som_read_data('330_2006-alle-labels-som.data');
sD131=som_read_data('131_var_all_labels_som.data');
sD131net= som_read_data('131_var_netz_som.data');
sDall = som_read_data('sorted_all-som.data');

% convert labels to strings instead of character arrays; transpose
% label = transpose(cellstr(label));


%% data-specific stuff
% generate three data sets for the nnet to play with
% one: N1, Yield2003, EM38 -- target: Yield2004
% two: N1, Yield2003, EM38, N2, REIP32 -- target: Yield2004
% three: N1, Yield2003, EM38, N2, REIP32, N3, REIP49 -- target: Yield2004
% works by eliminating the respective columns from the 'data' matrix above

%set_2(:,3) = [];

% only set_3 is actually used here
%set_3 = data;
%set_3(:,7) = [];
%label(:,1) = []; % delete "ID" label
%label(:,7) = [];


%Size_set_3 = size(set_3);
sDall.comp_names(8,:) = [];
sDall.comp_names(1,:) = [];
sDall.data(:,8) = [];
sDall.data(:,1) = [];

%%
%sDall.c

modelcomparison(sD330.data,'F330',20,sD330.comp_names);
modelcomparison(sD131.data,'F131',20,sD131.comp_names);
modelcomparison(sD131net.data,'F131net',20,sD131net.comp_names);
modelcomparison(sDall.data,'F04',20,sDall.comp_names);


%% get script stats
duration = etime(clock, clock_start)