tic
clear all
clc
[FileName,PathName,index]=uigetfile('*.xlsx','读取光谱','OriginalSpectr');
str=[PathName,FileName];
[data,txt]=xlsread(str,1);
x = data(:,1:end-1);
y = data(:,end);

% 训练集——20个样本
p_train = x(1:20,1:end);
%disp(p_train)
t_train = y(1:20,end);
Train_max_min=max(t_train)-min(t_train);

% 测试集——10个样本
p_test = x(21:end,1:end);
%disp(p_test)
t_test = y(21:end,end);
Test_max_min=max(t_test)-min(t_test);

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%构建多元回归模型
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%%%%%%%%%%%%%%%%%%%%%%%%%
%第一个模型：多元线性回归模型构建
%%%%%%%%%%%%%%%%%%%%%%%%%
%disp(p_test)
yzX=[ones(10,1),p_test];
X=[ones(20,1),p_train];
[b1,bint,r,rint,s]=regress(t_train,X);%进行多元线性回归，b为回归系数，其中第一个值为常数项
for i=1:20
    ymodel1_jm(i,1)=b1'*X(i,:)';%ymodel为模拟值，结果为模拟值与实测值的相关系数R，用于评价模型构建精度
end
r1=corrcoef(t_train,ymodel1_jm);
a1_R1 = r1(1,2)^2;%建模决定系数
a1_rmsejm = sqrt(sum((t_train'-ymodel1_jm').^2)/20); %模型构建的均方根误差
a1_nrmsejm = a1_rmsejm./Train_max_min;
%%%%%%%%%%%%%%%%%%%%%%%%%%
%多元线性回归的模型验证
%%%%%%%%%%%%%%%%%%%%%%%%%%
for i=1:10
    YMODEL1_yz(i,1)=b1'*yzX(i,:)';%YMODEL为预测值,用于模型精度验证
end
r2=corrcoef(t_test,YMODEL1_yz);
a1_R2 = r2(1,2)^2;%验证决定系数
a1_rmseyz = sqrt(sum((t_test'-YMODEL1_yz').^2)/10); %模型预测的均方根误差
a1_nrmseyz = a1_rmseyz./Test_max_min;

% Plot of Predicted vs Reference values
figure, hold on grid
plot(t_test,YMODEL1_yz,'o')
% plot(y,yhat,'o')
xlabel('Reference y value'),ylabel('Predicted y value')
h = gca; xlim = get(h,'xlim');
h = line(xlim,xlim);

title(['a1_rmsejm = ' num2str(a1_rmsejm) ', a1_R1 = ' num2str(a1_R1) ', a1_rmseyz  = ' num2str(a1_rmseyz ) ' ,a1_R2 = ' num2str(a1_R2)])