prueba_flores_lda.m

Implementation to linear, quadratic and logistic discriminant analysis, for examples

%function prueba_FLORES_LDA
clc,close all,clear all;
%Problema de 3 Clases y 4 variables
SETOSA=1;
VIRGINIC=2;
VERSICOL=3;
%Datos vinculados a las clases
Datos =[5.0 3.3   1.4  0.2  SETOSA;
        6.4  2.8  5.6  2.2  VIRGINIC;
        4.8  3.1  1.6  0.2  SETOSA;            
        6.5  2.8  4.6  1.5  VERSICOL;
        6.7  3.1  5.6  2.4  VIRGINIC;
        6.3  2.8  5.1  1.5  VIRGINIC;
        4.6  3.4  1.4  0.5  SETOSA;
        6.9  3.1  5.1  2.3  VIRGINIC;
        6.2  2.2  4.5  1.5  VERSICOL;
        5.9  3.2  4.8  1.8  VERSICOL;
        4.6  3.6  1.0  0.2  SETOSA;
        6.1  3.0  4.6  1.4  VERSICOL;
        6.0  2.7  5.1  1.6  VERSICOL;
        6.5  3.0  5.2  2.0  VIRGINIC;
        5.6  2.5  3.9  1.1  VERSICOL;
        6.5  3.0  5.5  1.8  VIRGINIC;
        5.8  2.7  5.1  1.9  VIRGINIC;
        6.8  3.2  5.9  2.3  VIRGINIC;
        5.1  3.3  1.7  0.5  SETOSA;
        5.7  2.8  4.5  1.3  VERSICOL;
        6.2  3.4  5.4  2.3  VIRGINIC;
        7.7  3.8  6.7  2.2  VIRGINIC;
        6.3  3.3  4.7  1.6  VERSICOL;
        6.7  3.3  5.7  2.5  VIRGINIC;
        7.6  3.0  6.6  2.1  VIRGINIC;
        4.9  2.5  4.5  1.7  VIRGINIC;
        5.5  3.5  1.3  0.2  SETOSA;
        6.7  3.0  5.2  2.3  VIRGINIC;
        7.0  3.2  4.7  1.4  VERSICOL;
        6.4  3.2  4.5  1.5  VERSICOL;
        6.1  2.8  4.0  1.3  VERSICOL;
        5.9  3.0  5.1  1.8  VIRGINIC];
%**************************************************************************
%                                  ENTRENAMIENTO
%**************************************************************************
%Datos de entrenamiento
Img1Entr  = Datos([1:22],[1:4]);
MsckEntr = Datos([1:22],5);
f = lda( Img1Entr, MsckEntr); 
g = qda( Img1Entr, MsckEntr);
h = logda( Img1Entr, MsckEntr);
whos
disp('_____________________LDA_________________');disp(f)
disp('_____________________QDA_________________');disp(g)
disp('____________________LOGDA________________');disp(h)
[lambdaF ratioF] = cvar(f);   % canonical variates
%[lambdaG ratioG] = cvar(g);   % canonical variates
%[lambdaH ratioH] = cvar(h);   % canonical variates
%**************************************************************************
%                                  VALIDACION
%**************************************************************************
%Datos de validacion
Img1Vald = Datos([23:end],[1:4]);
MsckVald = Datos([23:end],5);

%Utilizando un LDA
[MsckExpF postF] = classify(f, Img1Vald);
%Utilizando un QDA
[MsckExpG postG] = classify(g, Img1Vald);
%Utilizando un LOGDA
[MsckExpH postH] = classify(h, Img1Vald);

%comparacion de errores
[mFG alphaFG] = mcnemar(MsckVald, f(Img1Vald), g(Img1Vald));
[mFH alphaFH] = mcnemar(MsckVald, f(Img1Vald), h(Img1Vald));
[mHG alphaHG] = mcnemar(MsckVald, h(Img1Vald), g(Img1Vald));
fprintf('\n mFG =%f mFH =%f mHG =%f \n', mFG, mFH, mHG);
fprintf('\n alphaFG = %f alphaFH = %f alphaHG = %f \n', alphaFG, alphaFH, alphaHG);
%**************************************************************************
%                                  GRAFICAS
%**************************************************************************
fg1=figure(1);
set(fg1,'Color','w');
set(fg1,'Name','MATRIZ DE CONFUSION','NumberTitle','on');
subplot(131); matrizConfusion(MsckExpF(:),MsckVald(:)); title('LDA');
subplot(132); matrizConfusion(MsckExpG(:),MsckVald(:)); title('QDA');
subplot(133); matrizConfusion(MsckExpH(:),MsckVald(:)); title('LOGDA');