Algorithm 1: Method to Determine Condition Attribute Weights

Input: the knowledge base $K=(U, R)$, $R=C{\displaystyle \cap }D;$

Output: the weight of the condition attributes, $W_i(c)=\{w(c_1),\dots ,w(c_j),\dots ,w(c_n)\}$;

1:  compute the equivalence class ${\left[X\right]}_C={\{X}_1{,\dots ,X}_j{,\dots X}_n\}$, ${\left[X\right]}_D={\{D}_1{,\dots ,D}_i{,\dots ,D}_m\}$

2:  for i = 1 to m do 3:    compute the relative knowledge attribute discernibility ${\mathit{KFDisR}}_{D_i}(C)$

4:    for j = 1 to n do 5:       compute the equivalence class ${\left[X_j\right]}_{(C-c_j)}=U/(C-c_j)$

6:       compute the relative knowledge attribute discernibility ${\mathit{KFDisR}}_{D_i}(C-c_j)$

7:       compute the relative attribute importance of conditional attribute $c_j,$ ${\mathit{Sign}}_{D_i}(c_j)$

8:    end for

9:    computer the condition attribute weight $W_i(c)$

10:  end for