% Copyright (c) 2002,  KULeuven-ESAT-SCD, License & help @ http://www.esat.kuleuven.ac.be/sista/lssvmlabclc;disp(' This demo explains the use of the advanced object oriented interface');disp('  ''model''. For first users, we recommend the functional interface');disp('  as  explained in ''democlass'' and ''demofun''.');disp('  ');disp(' The ''model'' is the generic object which collects '); disp(' all relevant signals, parameters, options and functions ');disp(' related to an application of the LS-SVM.');disp(' This interface is depreciated for the casional users because');disp(' of the implicit nature: the distinction between in- and ');disp(' output vanishes. If one wants to use the full power  ');disp(' of LS-SVMlab, one is recommended to go through this demo.');disp(' ');disp(' We focus on function estimation, however the insights are completely');disp( 'equivalent for classification. A dataset is constructed at first:');disp(' ');disp('> X = (-3:.2:3)''');X = (-3:.2:3)';disp('> Y = X.^3+2.*randn(length(X),1);');Y = X.^3+2.*randn(length(X),1);disp(' ');disp(' A model is CONSTRUCTED for this data');disp(' ');disp(' >> gam=1; sig2=1;'); gam=1; sig2=1;disp(' >> model = initlssvm(X,Y,''function'',gam,sig2,''RBF_kernel'');');model = initlssvm(X,Y,'function',gam,sig2,'RBF_kernel');disp(' ');disp('press enter to continue...');pause; disp('  The specifications of the model can be seen typing just the'); disp('  name of the object that is constructed.');disp(' >> model');modeldisp(' ');disp(' If one wants to see the value of a specific option,');disp(' the ''.'' operator is to be used:');disp(' ');disp(' >> model.preprocess'); model.preprocessdisp(' ');disp('press enter to continue...');pause; disp(' If one wants to CHANGE a value of a specific option of the');disp(' model, the function ''changelssvm'' is to be used if consistent');disp(' models are wanted:');disp(' ');disp(' >> model = changelssvm(model,''implementation'',''MATLAB'');');model = changelssvm(model,'implementation','MATLAB');disp(' ');disp(' The options can be divided in 5 classes, the general LS-SVM options,');disp(' the trainpoint administration, the options of the Conjugate Gradient algorithm,');disp(' the preprocess options and the encoding options.');disp(' The help of ''changelssvm'' gives a description of the different');disp(' fields of the model (type ''help changelssvm''). ');disp(' The use of the model''s field ''status'' is important to see the full power');disp(' of the model concept. The object ''model'' knows ');disp(' whether it needs to be retrained. Retraining is needed if a specification');disp(' of the model is changed since the last training.');disp(' ');disp(' The demo ... shows how to control the preprocessing and the coding');disp(' using the appropriate model.');disp(' ');disp(' It is adviced to check carefully the model''s options before'); disp(' starting the calculations.');disp(' ');disp('press enter to continue...');pause; disp(' If the model is clearly defined, the routine to train the model');disp(' has to be called on this model. In the case of the LS-SVM, ');disp(' this training is done by:');disp(' ');disp(' >> model = trainlssvm(model);');model = trainlssvm(model);disp(' ');disp(' Given the trained model, one can simulate some testpoints');disp(' and make a plot of the model. (wait a few seconds...)');disp(' ');disp(' >> Xt = 2.*randn(10,1);');Xt = 2.*randn(10,1);disp(' >> Yt = simlssvm(model,Xt)');Yt= simlssvm(model,Xt)disp(' >> plotlssvm(model);');plotlssvm(model);disp(' ');disp(' By understanding this step, one masters basicly how to ');disp(' use the object ''model''.');disp(' ');disp(' As an extra, the underlying function is also given as a dotted line')disp(' on the plot');disp('>> hold on; plot((-3:.2:3)'', sinc(-3:.2:3)'','':'');');hold on; plot((-4:.2:4)', (-4:.2:4).^3','r:');disp(' ');disp('press enter to continue...');pause; disp(' The same model can be used to try different options:');disp(' ');disp(' >> model = changelssvm(model,''gam'',3);'); model = changelssvm(model,'gam',3);disp(' >> model = trainlssvm(model);'); model = trainlssvm(model);disp(' >> plotlssvm(model);');plotlssvm(model);disp(' ');disp(' or with a polynomial kernel of degree 3');disp(' ');disp(' >> model = changelssvm(model,''kernel_type'',''poly_kernel'');'); model = changelssvm(model,'kernel_type','poly_kernel');disp(' >> model = changelssvm(model,''kernel_pars'',[1;3]);');model = changelssvm(model,'kernel_pars',[1;3]);disp(' >> model = trainlssvm(model);'); model = trainlssvm(model);disp(' >> plotlssvm(model);');plotlssvm(model);disp(' ');disp(' This concludes this demo');