├── README.md
├── MFS-MCDM.m
└── Topsis.m


/README.md:
--------------------------------------------------------------------------------
1 | # MFS-MCDM-Multi-label-feature-selection-using-multi-criteria-decision-making-2020
2 | https://doi.org/10.1016/j.knosys.2020.106365
3 | Published in Knowledge-Based Systems journal
4 | # https://www.sciencedirect.com/science/article/abs/pii/S0950705120305116
5 | 


--------------------------------------------------------------------------------
/MFS-MCDM.m:
--------------------------------------------------------------------------------
 1 | 
 2 | % MFS-MCDM: Multi-label feature selection using multi-criteria decision making-2020
 3 | 
 4 | function S1= MFS_MCDM(dataTrain,dataTrainLabels)
 5 | 
 6 | [~, feature_num] = size( dataTrain);
 7 | Lambda = 10;
 8 | D = pinv(dataTrain'*dataTrain + Lambda*eye(feature_num)) * dataTrain'*dataTrainLabels;
 9 | for t = 1:num_class
10 |     w(t) = entropy(dataTrainLabels(:,t));
11 | end
12 | W = w./sum(w);
13 | rCloseness = Topsis(D,ones(1,feature_num),W);
14 | [SW S1] = sort(-rCloseness);
15 | end
16 | 


--------------------------------------------------------------------------------
/Topsis.m:
--------------------------------------------------------------------------------
 1 | function RC = Topsis(M,Redundancy,W)
 2 | %%%%%%%% step 1:normalization  %%%%%%%%%%%%%%%%%%
 3 | % M1=M.^2;
 4 | % M2=sqrt(sum(M1));
 5 |  [row column]=size(M);
 6 | % r=zeros(row,column);
 7 | % for i=1:row
 8 | %     for j=1:column
 9 | %         r(i,j)=M(i,j)/M2(j);%normalized decision matrix
10 | %     end
11 | % end
12 | r = M;
13 | %%%%%%%% step 2:Multiply Weighting  %%%%%%%%%%%%%%%%%%
14 | w1=zeros(row,column);
15 | for i=1:row
16 |     for j=1:column
17 |         w1(i,j)=r(i,j)*W(j);
18 |     end
19 | end 
20 | %%%%%%%%%%
21 | for i=1:row
22 |     for j=1:column
23 |         w1(i,j)=r(i,j)*Redundancy(i);
24 |     end
25 | end 
26 | %%%%%%%% step 3:Find ideal & negative ideal  %%%%%%%%%%%%%%%%%%
27 | ideal=max(w1);
28 | nideal=min(w1);
29 | 
30 | %%%%%%%% step 4:Calculate seperation from ideal & negative ideal  %%%%%%%%%%%%%%%%%%
31 | AP=zeros(row,column);
32 | for i=1:row
33 |     for j=1:column
34 |         AP(i,j)=(w1(i,j)-ideal(j))^2;
35 |     end
36 | end
37 | A1=sqrt(sum(AP, 2));
38 | AN=zeros(row,column);
39 | for i=1:row
40 |     for j=1:column
41 |         AN(i,j)=(w1(i,j)-nideal(j))^2;
42 |     end
43 | end
44 | A2=sqrt(sum(AN, 2));
45 | %%%%%%%% step 5:Calculate Relative Closness  %%%%%%%%%%%%%%%%%%
46 | RC=zeros(1,row);
47 |  for i=1:row
48 |      RC(i)=A2(i)/(A1(i)+A2(i));
49 |  end
50 |  
51 | end
52 | 


--------------------------------------------------------------------------------