Skip to main content
. 2021 Dec 1;23(12):1617. doi: 10.3390/e23121617
Algorithm 1. TCRFS.
  • Input: 

     

  •  

       A training sample D with a full feature set F={f1,f2,…,fn} and the label set L={l1,l2,…,lm}; User-specified threshold K.

  • Output: 

     

  •  
       The selected feature subset S.
    • 1:
      S←⌀;
    • 2:
      k←0;
    • 3:
      fori=1 to n do
    • 4:
         Calculate the feature relevance I(fi;li|lj);
    • 5:
      end for
    • 6:
      whilek<Kdo
    • 7:
         if k == 0 then
    • 8:
             Select the first feature fj with the largest I(fi;li|lj);
    • 9:
             k=k+1;
    • 10:
           S=S∪{fj};
    • 11:
           F=F−{fj};
    • 12:
         end if
    • 13:
         for each candidate feature fi∈F do
    • 14:
           According to the Formula (21) and calculate the J(fi);
    • 15:
         end for
    • 16:
         Select the feature fj with the largest J(fi);
    • 17:
         S=S∪{fj};
    • 18:
         F=F−{fj};
    • 19:
         k=k+1;
    • 20:
      end while