机器学习笔记关于MATLAB实现决策树

%% I. 清空环境变量
clear all
clc
warning off

%% II. 导入数据
load data.mat

%%
% 1. 随机产生训练集/测试集
a = randperm(569);
Train = data(a(1:500),:);
Test = data(a(501:end),:);

%%
% 2. 训练数据
P_train = Train(:,3:end);
T_train = Train(:,2);

%%
% 3. 测试数据
P_test = Test(:,3:end);
T_test = Test(:,2);

%% III. 创建决策树分类器
ctree = ClassificationTree.fit(P_train,T_train);

%%
% 1. 查看决策树视图
view(ctree);
view(ctree,‘mode‘,‘graph‘);

%% IV. 仿真测试
T_sim = predict(ctree,P_test);

%% V. 结果分析
count_B = length(find(T_train == 1));
count_M = length(find(T_train == 2));
rate_B = count_B / 500;
rate_M = count_M / 500;
total_B = length(find(data(:,2) == 1));
total_M = length(find(data(:,2) == 2));
number_B = length(find(T_test == 1));
number_M = length(find(T_test == 2));
number_B_sim = length(find(T_sim == 1 & T_test == 1));
number_M_sim = length(find(T_sim == 2 & T_test == 2));
disp([‘病例总数：‘ num2str(569)...
      ‘  良性：‘ num2str(total_B)...
      ‘  恶性：‘ num2str(total_M)]);
disp([‘训练集病例总数：‘ num2str(500)...
      ‘  良性：‘ num2str(count_B)...
      ‘  恶性：‘ num2str(count_M)]);
disp([‘测试集病例总数：‘ num2str(69)...
      ‘  良性：‘ num2str(number_B)...
      ‘  恶性：‘ num2str(number_M)]);
disp([‘良性乳腺肿瘤确诊：‘ num2str(number_B_sim)...
      ‘  误诊：‘ num2str(number_B - number_B_sim)...
      ‘  确诊率p1=‘ num2str(number_B_sim/number_B*100) ‘%‘]);
disp([‘恶性乳腺肿瘤确诊：‘ num2str(number_M_sim)...
      ‘  误诊：‘ num2str(number_M - number_M_sim)...
      ‘  确诊率p2=‘ num2str(number_M_sim/number_M*100) ‘%‘]);
  
%% VI. 叶子节点含有的最小样本数对决策树性能的影响
leafs = logspace(1,2,10);

N = numel(leafs);

err = zeros(N,1);
for n = 1:N
    t = ClassificationTree.fit(P_train,T_train,‘crossval‘,‘on‘,‘minleaf‘,leafs(n));
    err(n) = kfoldLoss(t);
end
plot(leafs,err);
xlabel(‘叶子节点含有的最小样本数‘);
ylabel(‘交叉验证误差‘);
title(‘叶子节点含有的最小样本数对决策树性能的影响‘)

%% VII. 设置minleaf为13，产生优化决策树
OptimalTree = ClassificationTree.fit(P_train,T_train,‘minleaf‘,13);
view(OptimalTree,‘mode‘,‘graph‘)

%%
% 1. 计算优化后决策树的重采样误差和交叉验证误差
resubOpt = resubLoss(OptimalTree)
lossOpt = kfoldLoss(crossval(OptimalTree))

%%
% 2. 计算优化前决策树的重采样误差和交叉验证误差
resubDefault = resubLoss(ctree)
lossDefault = kfoldLoss(crossval(ctree))

%% VIII. 剪枝
[~,~,~,bestlevel] = cvLoss(ctree,‘subtrees‘,‘all‘,‘treesize‘,‘min‘)
cptree = prune(ctree,‘Level‘,bestlevel);
view(cptree,‘mode‘,‘graph‘)

%%
% 1. 计算剪枝后决策树的重采样误差和交叉验证误差
resubPrune = resubLoss(cptree)
lossPrune = kfoldLoss(crossval(cptree))

Decision tree for classification
 1  if x23<112.8 then node 2 elseif x23>=112.8 then node 3 else 1
 2  if x28<0.1456 then node 4 elseif x28>=0.1456 then node 5 else 1
 3  if x7<0.07214 then node 6 elseif x7>=0.07214 then node 7 else 2
 4  if x28<0.1358 then node 8 elseif x28>=0.1358 then node 9 else 1
 5  if x22<23.74 then node 10 elseif x22>=23.74 then node 11 else 2
 6  if x2<19.83 then node 12 elseif x2>=19.83 then node 13 else 2
 7  if x8<0.03456 then node 14 elseif x8>=0.03456 then node 15 else 2
 8  if x11<1.04755 then node 16 elseif x11>=1.04755 then node 17 else 1
 9  if x28<0.139 then node 18 elseif x28>=0.139 then node 19 else 1
10  class = 1
11  if x5<0.09096 then node 20 elseif x5>=0.09096 then node 21 else 2
12  class = 1
13  class = 2
14  class = 1
15  class = 2
16  if x21<16.805 then node 22 elseif x21>=16.805 then node 23 else 1
17  class = 2
18  class = 2
19  class = 1
20  class = 1
21  class = 2
22  if x14<48.975 then node 24 elseif x14>=48.975 then node 25 else 1
23  class = 1
24  if x15<0.003294 then node 26 elseif x15>=0.003294 then node 27 else 1
25  class = 1
26  class = 1
27  if x22<33.35 then node 28 elseif x22>=33.35 then node 29 else 1
28  class = 1
29  if x22<33.56 then node 30 elseif x22>=33.56 then node 31 else 1
30  class = 2
31  class = 1

病例总数：569  良性：357  恶性：212
训练集病例总数：500  良性：310  恶性：190
测试集病例总数：69  良性：47  恶性：22
良性乳腺肿瘤确诊：45  误诊：2  确诊率p1=95.7447%
恶性乳腺肿瘤确诊：19  误诊：3  确诊率p2=86.3636%

resubOpt =

    0.0460


lossOpt =

    0.0740


resubDefault =

    0.0140


lossDefault =

    0.0820


bestlevel =

     4


resubPrune =

    0.0300


lossPrune =

    0.0800