山东大学机器学习（实验四解读）——朴素贝叶斯

实验所给的convData.m是将大小为12960的nursery.data.txt数据集分割成一份大小为10000的训练集training_data.txt和大小为2960的测试集test_data.txt。

问题1

%--------------------------------LogMLE.m--------------------------------
function [label_pre,sum,sucess_rate] = LogMLE(test,train)
    [train_row,train_col] = size(train);
    test_num = size(test,1);
    %记录每个类对应的对数似然函数值，最大值对应的下标减1即最终我们预测的类
    label_for_MLE = zeros(test_num,5);
    label_pre = zeros(test_num,1); %存储所有我们预测的类
    label_num = max(train(:,end)); %类别总数
    for i = 1:test_num
        for y = 0:label_num
            count_y = length(find(train(:,train_col)==y));
            p_y = count_y/(train_row+label_num);
            log_count_xy = 0;
            for j = 1:train_col-1
                count_xy =