MATLAB编程实现西瓜集2.0和信息增益选择最优划分特征构建决策树，写出完整代码

MATLAB中构建决策树通常使用`treebagger`函数，这个函数可以自动选择最优特征来进行划分。但是为了实现西瓜集2.0（一种数据集模拟决策过程）并结合信息增益，我们需要手动编写一些辅助函数来计算信息增益。以下是一个简化版的示例代码，假设西瓜集2.0的数据已经准备好了：

% 西瓜集2.0数据，这里仅做演示，实际数据需要替换
data = readtable('西瓜数据.csv'); % 假设数据来自CSV文件
X = data(:, 1:end-1); % 特征
Y = data(:, end); % 类别

% 计算信息增益
function ig = calculate_info_gain(X, Y, feature)
    num_classes = unique(Y);
    entropy_Y = -sum(Y) * log(sum(Y) / length(Y));
    
    sub_data = cellfun(@(x) x(:, feature), partition(X, Y), 'UniformOutput', false);
    info_gains = zeros(size(feature));
    
    for i = 1:length(num_classes)
        sub_entropy = zeros(length(feature), 1);
        for j = 1:length(sub_data)
            sub_class = sub_data{j}(:, feature);
            sub_entropy(j) = -sum(sub_class == num_classes(i)) * log(sum(sub_class == num_classes(i)) / sum(sub_data{j}));
        end
        ig_i = entropy_Y - mean(sub_entropy);
        info_gains(i) = ig_i;
    end
    ig = max(info_gains);
end

% 构建决策树，这里使用递归算法简单模拟，真实应用应使用treebagger
function tree = build_decision_tree(X, Y, depth, min_samples_split)
    if depth == 0 || size(X, 1) < min_samples_split || all(Y == Y(1))
        % 如果达到深度限制或样本太少，直接返回类别
        tree.classifier = mode(Y);
        return
    end
    
    % 找到信息增益最大的特征
    [~, best_feature] = max(calculate_info_gain(X, Y, 1:size(X, 2)));
    
    % 划分数据
    [left_X, right_X, left_Y, right_Y] = split_dataset(X, Y, best_feature, find(X(:, best_feature) == 1));
    
    % 递归构建子树
    tree.left = build_decision_tree(left_X, left_Y, depth - 1, min_samples_split);
    tree.right = build_decision_tree(right_X, right_Y, depth - 1, min_samples_split);
    
    % 存储特征和阈值
    tree.feature = best_feature;
    tree.threshold = X(find(X(:, best_feature) == 1), best_feature);
end

% 分割函数
function [left, right, left_Y, right_Y] = split_dataset(X, Y, feature, condition)
    left = X(condition, :);
    right = X(~condition, :);
    left_Y = Y(condition);
    right_Y = Y(~condition);
end

% 主程序调用
min_samples_split = 5; % 设置最小样本数分裂节点
depth = 3; % 设定最大深度
tree = build_decision_tree(X, Y, depth, min_samples_split);

% 输出结果
disp(tree);