%%用最近鄰方法分類    采用歐氏距離
function [class,mis_num,correct_rate]=classify_NN(data,class_num,train_num,dis_mark)
[ dim,total_num ] = size( data ) ;
class_data_num = total_num/class_num ;   %得到每類中的樣本數目
for class_mark = 1:class_num                   %for循環得到用來訓練的數據
    for class_data_mark = 1:train_num
      train_data( :,( class_mark-1 )*train_num+class_data_mark ) = data( :,( class_mark-1 )*class_data_num+class_data_mark );
      %test_data(:,())
    end
end
test_num = class_data_num-train_num;
for class_mark = 1:class_num                   %for循環得到用來測試的數據
    for class_data_mark = 1:test_num
     test_data( :,( class_mark-1 )*test_num+class_data_mark ) = data( :,( class_mark-1 )*class_data_num+class_data_mark+train_num );
      %test_data(:,())
    end
end
train_data_num = train_num*class_num;
test_data_num = test_num*class_num;
for test_data_mark = 1:test_data_num                %for循環求測試數據與每個訓練數據的距離
    for train_data_mark = 1:train_data_num
        switch dis_mark
            case  'L2'
        dis(test_data_mark,train_data_mark) = norm(test_data(:,test_data_mark)-train_data(:,train_data_mark));
            case  'cos'
         dis(test_data_mark,train_data_mark) = acos((test_data(:,test_data_mark)'*train_data(:,train_data_mark))/(norm(test_data(:,test_data_mark))*norm(train_data(:,train_data_mark))));
            case  'L1'
         dis(test_data_mark,train_data_mark) =  norm(test_data(:,test_data_mark)-train_data(:,train_data_mark),1);
            otherwise
         dis(test_data_mark,train_data_mark) = norm(test_data(:,test_data_mark)-train_data(:,train_data_mark));       
        end
    end
end
[mindis,index]=min(dis');   
class = fix((index+train_num-1)/train_num);    %求測試數據所屬的類別  
mis_num = 0;
for test_data_mark = 1:test_data_num        
    if class(test_data_mark)~=fix((test_data_mark+test_num-1)/test_num)   %若算得的類別與實際所屬類別不一樣則分類錯誤
        mis_num = mis_num+1;
    end
end
correct_rate =(test_data_num-mis_num)/test_data_num;