function untitled() load('D:\mat格式的MNIST数据\test_labels.mat') load('D:\mat格式的MNIST数据\train_images.mat') load('D:\mat格式的MNIST数据\train_labels.mat') load('D:\mat格式的MNIST数据\test_images.mat') train_num = 600; test_num = 200; %训练数据,图像转向量 data_train = mat2vector(train_images(:,:,1:train_num),train_num); data_test = mat2vector(test_images(:,:,1:test_num),test_num); % 处理训练数据,防止后验概率为0 [data_train,position] = fun(data_train,train_labels1(1:train_num)'); % 处理测试数据 for rows = 1:10 data_test(:,position{1,rows})=[]; end %模型部分 Mdl = fitcnb(data_train,train_labels1(1:train_num)); %测试结果 result = predict(Mdl,data_test); result = result.'; xlabel=[0,1,2,3,4,5,6,7,8,9]; resultbar = [0,0,0,0,0,0,0,0,0,0]; testbar = [0,0,0,0,0,0,0,0,0,0]; for i = 1:test_num temp1=result(i); temp1=temp1+1; resultbar(temp1)=resultbar(temp1)+1; temp2=test_labels1(i); temp2=temp2+1; testbar(temp2)=testbar(temp2)+1; end bar(xlabel, [resultbar' testbar']); % 整体正确率 acc = 0.; for i = 1:test_num if result(i)==test_labels1(i) acc = acc+1; end end title('精确度为：',(acc/test_num)*100) end function [output,position] = fun(data,label) position = cell(1,10); %创建cell存储每类中删除的列标 for i = 0:9 temp = []; pos = []; for rows = 1:size(data,1) if label(rows)==i temp = [temp;data(rows,:)]; end end for cols = 1:size(temp,2) var_data = var(temp(:,cols)); if var_data==0 pos = [pos,cols]; end end position{i+1} = pos; data(:,pos)=[]; end output = data; end function [data_]= mat2vector(data,num) [row,col,~] = size(data); data_ = zeros(num,row*col); for page = 1:num for rows = 1:row for cols = 1:col data_(page,((rows-1)*col+cols)) = im2double(data(rows,cols,page)); end end end end 将画图部分重写，完成相同功能

好的，以下是将画图部分重写的代码：

% 统计预测结果和真实结果的数量
result_count = histcounts(result, 0:10);
test_count = histcounts(test_labels1, 0:10);

% 绘制柱状图
figure;
bar(0:9, [result_count', test_count']);
xlabel('数字');
ylabel('数量');
legend('预测结果', '真实结果');

% 计算并输出整体正确率
accuracy = sum(result == test_labels1) / test_num;
title(['整体正确率为：', num2str(accuracy * 100), '%']);

改动后，完整代码如下：

function untitled()
    load('D:\mat格式的MNIST数据\test_labels.mat')
    load('D:\mat格式的MNIST数据\train_images.mat')
    load('D:\mat格式的MNIST数据\train_labels.mat')
    load('D:\mat格式的MNIST数据\test_images.mat')
    
    train_num = 600;
    test_num = 200;
    
    % 训练数据,图像转向量
    data_train = mat2vector(train_images(:,:,1:train_num),train_num);
    data_test = mat2vector(test_images(:,:,1:test_num),test_num);
    
    % 处理训练数据,防止后验概率为0
    [data_train, position] = fun(data_train,train_labels1(1:train_num)');
    
    % 处理测试数据
    for rows = 1:10
        data_test(:,position{1,rows})=[];
    end
    
    % 模型部分
    Mdl = fitcnb(data_train,train_labels1(1:train_num));
    
    % 测试结果
    result = predict(Mdl,data_test);
    result = result.';
    
    % 统计预测结果和真实结果的数量
    result_count = histcounts(result, 0:10);
    test_count = histcounts(test_labels1, 0:10);
    
    % 绘制柱状图
    figure;
    bar(0:9, [result_count', test_count']);
    xlabel('数字');
    ylabel('数量');
    legend('预测结果', '真实结果');
    
    % 计算并输出整体正确率
    accuracy = sum(result == test_labels1) / test_num;
    title(['整体正确率为：', num2str(accuracy * 100), '%']);
end

function [output,position] = fun(data,label)
    position = cell(1,10); % 创建cell存储每类中删除的列标
    for i = 0:9
        temp = [];
        pos = [];
        for rows = 1:size(data,1)
            if label(rows)==i
                temp = [temp;data(rows,:)];
            end
        end
        for cols = 1:size(temp,2)
            var_data = var(temp(:,cols));
            if var_data==0
                pos = [pos,cols];
            end
        end
        position{i+1} = pos;
        data(:,pos)=[];
    end
    output = data;
end

function [data_] = mat2vector(data,num)
    [row,col,~] = size(data);
    data_ = zeros(num,row*col);
    for page = 1:num
        for rows = 1:row
            for cols = 1:col
                data_(page,((rows-1)*col+cols)) = im2double(data(rows,cols,page));
            end
        end
    end
end