Parallel Attribute Reduction Algorithm for Complex Heterogeneous Data Using MapReduce

Algorithm 7

PARA_NMG algorithm.

Input: ,

Output: reduction reduct

 begin

 Step 1: initialize , , ;

 Step 2: if (C-reduct)= , go to Step 3; while if (C-reduct) , execute the following loop operations:

  Step 2.1: for each condition attribute , use algorithms 1~4 to calculate the positive region POS_k of ;

  Step 2.2: compare the positive region POS_k of condition attribute subset after each attribute c_k added. Find current maximum positive region Max_POS; if then keep attribute reduction invariant, and go to Step 3. Otherwise, add c_k into reduct, ;

  Step 2.3: update boundary sample set Q with algorithms 5~6, , and return to Step 2.1

 Step 3: output reduction reduct

 end.