用matlab代码实现朴素贝叶斯分类器，要求如下：1.导入鸢尾花数据集； 2.分别取三个类中每个类的40个样本作为训练集，10个样本作为测试集； 3.计算每个类的概率密度函数，即每个类的样本均值与方差（假设样本特征数据满足高斯分布）； 4.将测试样本带入每个类的概率密度函数，并判断该样本属于哪个类别； 5.计算精确率，召回率，F度量

好的，我可以帮你实现这个问题。下面是实现步骤：

1. 导入鸢尾花数据集

load fisheriris
X = meas; % 特征数据
Y = species; % 标签数据

2. 分别取三个类中每个类的40个样本作为训练集，10个样本作为测试集

train_data = [];
train_label = [];
test_data = [];
test_label = [];
for i = 1:3
    idx = find(strcmp(Y, unique(Y(i))));
    idx_train = idx(1:40);
    idx_test = idx(41:50);
    train_data = [train_data; X(idx_train, :)];
    train_label = [train_label; Y(idx_train)];
    test_data = [test_data; X(idx_test, :)];
    test_label = [test_label; Y(idx_test)];
end

3. 计算每个类的概率密度函数，即每个类的样本均值与方差

% 计算每个类的均值和方差
mu = zeros(3, size(X, 2));
sigma = zeros(3, size(X, 2));
for i = 1:3
    idx = find(strcmp(train_label, unique(Y(i))));
    mu(i, :) = mean(train_data(idx, :));
    sigma(i, :) = var(train_data(idx, :));
end

4. 将测试样本带入每个类的概率密度函数，并判断该样本属于哪个类别

% 计算测试集的预测标签
pred = zeros(size(test_data, 1), 1);
for i = 1:size(test_data, 1)
    prob = zeros(3, 1);
    for j = 1:3
        p = 1;
        for k = 1:size(test_data, 2)
            p = p * normpdf(test_data(i, k), mu(j, k), sqrt(sigma(j, k)));
        end
        prob(j) = p;
    end
    [~, idx] = max(prob);
    pred(i) = idx;
end

5. 计算精确率，召回率，F度量

% 计算评价指标
accuracy = sum(pred == test_label) / length(test_label);
precision = zeros(3, 1);
recall = zeros(3, 1);
F_measure = zeros(3, 1);
for i = 1:3
    idx = find(strcmp(test_label, unique(Y(i))));
    TP = sum(pred(idx) == i);
    FP = sum(pred(idx) ~= i);
    FN = sum(test_label(idx) ~= i);
    precision(i) = TP / (TP + FP);
    recall(i) = TP / (TP + FN);
    F_measure(i) = 2 * precision(i) * recall(i) / (precision(i) + recall(i));
end

完整代码如下：

load fisheriris
X = meas;
Y = species;

% 分别取三个类中每个类的40个样本作为训练集，10个样本作为测试集
train_data = [];
train_label = [];
test_data = [];
test_label = [];
for i = 1:3
    idx = find(strcmp(Y, unique(Y(i))));
    idx_train = idx(1:40);
    idx_test = idx(41:50);
    train_data = [train_data; X(idx_train, :)];
    train_label = [train_label; Y(idx_train)];
    test_data = [test_data; X(idx_test, :)];
    test_label = [test_label; Y(idx_test)];
end

% 计算每个类的均值和方差
mu = zeros(3, size(X, 2));
sigma = zeros(3, size(X, 2));
for i = 1:3
    idx = find(strcmp(train_label, unique(Y(i))));
    mu(i, :) = mean(train_data(idx, :));
    sigma(i, :) = var(train_data(idx, :));
end

% 计算测试集的预测标签
pred = zeros(size(test_data, 1), 1);
for i = 1:size(test_data, 1)
    prob = zeros(3, 1);
    for j = 1:3
        p = 1;
        for k = 1:size(test_data, 2)
            p = p * normpdf(test_data(i, k), mu(j, k), sqrt(sigma(j, k)));
        end
        prob(j) = p;
    end
    [~, idx] = max(prob);
    pred(i) = idx;
end

% 计算评价指标
accuracy = sum(pred == test_label) / length(test_label);
precision = zeros(3, 1);
recall = zeros(3, 1);
F_measure = zeros(3, 1);
for i = 1:3
    idx = find(strcmp(test_label, unique(Y(i))));
    TP = sum(pred(idx) == i);
    FP = sum(pred(idx) ~= i);
    FN = sum(test_label(idx) ~= i);
    precision(i) = TP / (TP + FP);
    recall(i) = TP / (TP + FN);
    F_measure(i) = 2 * precision(i) * recall(i) / (precision(i) + recall(i));
end