MATLAB code for machine learning--random forest (1)

code:

%% 初始化数据
clc
clear
close all
%%  导入数据
data = xlsread('数据集.xlsx','Sheet1','A1:F100');%导入数据库

%%  划分训练集和测试集
TE= randperm(100);%将数据打乱，重新排序；

PN = data(TE(1: 80), 1: 5)';%划分训练集输入
TN = data(TE(1: 80), 6)';%划分训练集输出

PM = data(TE(81: end), 1: 5)';%划分测试集输入
TM = data(TE(81: end), 6)';%划分测试集输出

%%  数据归一化
[pn, ps_input] = mapminmax(PN, 0, 1);%归一化到（0，1）
pn=pn';
pm = mapminmax('apply', PM, ps_input);%引用结构体，保持归一化方法一致；
pm=pm';
[tn, ps_output] = mapminmax(TN, 0, 1);
tn=tn';

%%  模型参数设置及训练模型
trees = 100; % 决策树数目
leaf  = 5; % 最小叶子数
OOBPrediction = 'on';  % 打开误差图
OOBPredictorImportance = 'on'; % 计算特征重要性
Method = 'regression';  % 选择回归或分类
net = TreeBagger(trees, pn, tn, 'OOBPredictorImportance', OOBPredictorImportance,...
      'Method', Method, 'OOBPrediction', OOBPrediction, 'minleaf', leaf);
importance = net.OOBPermutedPredictorDeltaError;  % 重要性

%%  仿真测试
pyuce = predict(net, pm );

%%  数据反归一化
Pyuce = mapminmax('reverse', pyuce, ps_output);
Pyuce =Pyuce';

%%  绘图
figure %画图真实值与预测值对比图
plot(TM,'bo-')
hold on
plot(Pyuce,'r*-')
hold on
legend('真实值','预测值')
xlabel('预测样本')
ylabel('预测结果')
grid  on

figure % 绘制特征重要性图
bar(importance)
legend('各因素重要性')
xlabel('特征')
ylabel('重要性')

%%  相关指标计算
error=Pyuce-TM;
[~,len]=size(TM);
R2=1-sum((TM-Pyuce).^2)/sum((mean(TM)-TM).^2);%相关性系数
MSE=error*error'/len;%均方误差
RMSE=MSE^(1/2);%均方根误差
disp(['测试集数据的MSE为：', num2str(MSE)])
disp(['测试集数据的MBE为：', num2str(RMSE)])
disp(['测试集数据的R2为：', num2str(R2)])

screenshot of data

Result:



If you need data and code compression package, please send an email message in the comment area, and it will be sent within one day, remember to pay attention and like it! !