-
Notifications
You must be signed in to change notification settings - Fork 1
/
EODE.m
149 lines (141 loc) · 5.84 KB
/
EODE.m
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
numRun = 1;
for i=1:numRun
addpath(genpath('data'));
% Problem = {'PanCancer'};%
% Problem = {'Alizadeh-2000-v1'};
Problem = {'Alizadeh-2000-v1','Alizadeh-2000-v2','Alizadeh-2000-v3','Armstrong-2002-v1','Armstrong-2002-v2','Bhattacharjee-2001','Bittner-2000',...
'Bredel-2005','Chen-2002','Chowdary-2006','Dyrskjot-2003','Garber-2001','Golub-1999-v1','Golub-1999-v2','Gordon-2002','Khan-2001_database',...
'Laiho-2007_database','Lapointe-2004-v1','Lapointe-2004-v2','Liang-2005','Nutt-2003-v1','Nutt-2003-v2','Nutt-2003-v3','Pomeroy-2002-v1'...
,'Pomeroy-2002-v2','Ramaswamy-2001_database','Risinger-2003','Shipp-2002-v1','Singh-2002','Su-2001','Tomlins-2006-v1','Tomlins-2006-v2',...
'West-2001','Yeoh-2002-v1','Yeoh-2002-v2'};
% Model SETTINGS
params.numOfFolds = 5; % Create CROSS VALIDATION FOLDS,'DT','ANN'};%};%;};%,'NB','RF''DISCR','SVM','NB'
params.classifiers = {'DISCR','DT','KNN','ANN','SVM','NB'};
%% MAIN LOOP
parfor j = 1:length(Problem)
for eachClass = 1:1
p_name = Problem{j};
results = Training(p_name, params);
results.p_name = p_name;
saveResults(results);
end
end
end
function results = Training(p_name , params)
warning('off','all');
params.p_name = p_name;
traindata = load(['C:\Users\qywxb\Desktop\EODE\TrainData\',p_name]);
traindata = traindata.TrainData;
% traindata = traindata.traindata;
% traindata = traindata.train;
testdata = load(['C:\Users\qywxb\Desktop\EODE\TestData\',p_name]);
testdata = testdata.TestData;
% testdata = testdata.testdata;
% testdata = testdata.test;
data = traindata;
X = data(:,1:end-1);
y = data(:, end);
feat = data(:,1:end-1);
label = data(:,end);
%% select the best classifier for feature selection
indx = cvpartition(label,'KFold',5);
for f = 1 : 5
test = data(indx.test(f),:);
train = data(~indx.test(f),:);
xtrain = train(:,1:end-1);
ytrain = train(:,end);
xtest = test(:,1:end-1);
ytest = test(:,end);
My_Model1 = fitcdiscr(xtrain,ytrain, 'discrimtype','diaglinear'); % DISCR
pred1 = predict(My_Model1,xtest);
acc1(f) = mean(pred1 == ytest);
My_Model2 = fitctree(xtrain,ytrain); % DT
pred2 = predict(My_Model2,xtest);
acc2(f) = mean(pred2 == ytest);
My_Model3 = fitcknn(xtrain,ytrain,'NumNeighbors',3); % KNN
pred3 = predict(My_Model3,xtest);
acc3(f) = mean(pred3 == ytest);
My_Model4 = trainNN(xtrain,ytrain); % ANN
pred4 = getNNPredict(My_Model4,xtest);
acc4(f) = mean(pred4 == ytest);
radial=templateSVM('KernelFunction','rbf','IterationLimit',50000,'Standardize',true); % SVM
My_Model5 = fitcecoc(xtrain,ytrain, 'learners', radial, 'ClassNames',[unique(ytrain)]);
pred5 = predict(My_Model5,xtest);
acc5(f) = mean(pred5 == ytest);
My_Model6 = fitcnb(xtrain,ytrain, 'distribution', 'kernel'); % NB
pred6 = predict(My_Model6,xtest);
acc6(f) = mean(pred6 == ytest);
end
disp([mean(acc1),mean(acc2),mean(acc3),mean(acc4),mean(acc5),mean(acc6)]);
[Acc1, num] = max([mean(acc1),mean(acc2),mean(acc3),mean(acc4),mean(acc5),mean(acc6)]);
fprintf('\n The %d th classifier is selected',num);
%% basic settings of GWO algorithm
% Number of k in K-nearest neighbor
opts.k = 3;
% Common parameter settings
opts.N = 100; % number of solutions
opts.T = 50; % maximum number of iterations
opts.num = num;
%% feature selection using GWO algorithm
GWO = jGreyWolfOptimizer(feat,label,opts);
sf = GWO.sf;
GWOdata = [feat(:,sf),label];
traindata = GWOdata;
numSF = size(traindata, 2) - 1 ;
feature = testdata(:,sf);
% save('sf.mat','sf');
testdata = [feature, testdata(:,end)];
finalClassifiers = [];
cvs = cvpartition(traindata(:,end),'KFold',5);
for f = 1 : 5
idxs = cvs.test(f);
testData = traindata(idxs,:);
trainData = traindata(~idxs,:);
%% Ensemble classifier for classification
train = trainData;
test = testData;
allClusters = generateClusters(train);
for i = 1 : length(allClusters)
balancedClusters{i} = allClusters{i};
end
classifierIndex = 1;
for c=balancedClusters
X = c{1,1}(:, 1:end-1);
y = c{1,1}(:, end);
all = trainClassifiers(X, y, params);
if size(all,1) < 1
continue
end
for temp = 1:length(all)
allclassifiers{classifierIndex} = all{1,temp};
classifierIndex = classifierIndex + 1;
end
end
[~, pred] = fusion(allclassifiers,test);
SC = [];
acc = zeros(1,size(pred,2));
for i = 1 : size(pred,2)
acc(i) = mean(pred(:,i) == test(:,end));
end
for i = 1 : size(pred,2)
if acc(i) >= mean(acc)
SC = [SC, i];
end
end
allclassifiers = allclassifiers(:,SC);
GWO = classifierSelectionGWO(allclassifiers, test, opts);
optimized_Accuracy(f) = fusion(allclassifiers(:,GWO.sc), testData);
nonoptimized_Accuracy(f) = fusion(allclassifiers, testData);
finalClassifiers = [finalClassifiers, allclassifiers(:,GWO.sc)];
end
% save('finalClassifiers.mat','finalClassifiers');
[testAcc, ~] = fusion(finalClassifiers, testdata);
results.testAcc = testAcc;
results.optimized_Accuracy = mean(optimized_Accuracy);
results.nonoptimized_Accuracy = mean(nonoptimized_Accuracy);
results.selected_Features = numSF;
disp(results.optimized_Accuracy);
disp(results.nonoptimized_Accuracy);
disp(numSF);
disp(results.testAcc);
end