Feature Selection Method Based on Artificial Bee Colony Algorithm and Support Vector Machines for Medical Datasets Classification

Pseudocode 3

Pseudo-code of developed feature selection algorithm based on ABC.

 𝑛= sample_count;

 Selected_features = {Ø}

 For 𝑝=1 to desired_feature_count

   For 𝑐=1 to feature_count

      Data_all = data(Selected_features + feature(c));

      For 𝑖=1 to class_number

     Train_data_class(i) = partition(rand(Data_all(class == i)), 0.75)

     Test_data_class(i) = partition(rand(Data_all(class = i)), others);

         [foods(i)] = Modified_ABC_algortihm (train_data_class(i), performance_function);

      End for 𝑖

   Test_data_all = merge(test_data_class);

   For 𝑖=1: size(Test_data_all)

      For 𝑘=1: class_count

    For 𝑗=1: count(foods(k))

    distance(k, j) = oklid_distance(foods(j, k)-test_data(i));

    End for 𝑗

    min_dist(k) = min(distance(k));

      End for 𝑘

   [result_of_test_data(i), class_of_test_data(i)] = min(min_dist);

   End for 𝑖

   Performance_criteria(feature(c))

= sum(class_of_test_data(i) == class_of_test_data_expected);

   End for c

 Best_feature(c) = arg max(performance_criteria(feature(c))

 Selected_features = Selected_fetures + best_feature(c)

 End for 𝑝