你好，我想预测未来(2018-2038年)的消费价值。但是我猜我的代码出了问题。这是消费的时间序列。我可以拟合真实样本和预测样本的高得分(如%99)。但是当我走到最后(预测部分)，在接近2或4步后，值开始是相同的。我认为这是正常化，因为消费必须增加，因为它的过去和最大值是1，所以值不能进一步，它保持在1以下。当我使用另一个代码的规范化，如Xn=规范化(Xn)网络找不到任何相关(NaN)。
除此之外，当我用这种方法进行标准化时，我如何去标准化样本。训练后的最大值和最小值是什么?我正在写下面的代码。提前谢谢。
十五=负载(“elektriktek.txt”);
xn =十五';
xn = (xn-min (xn)) / ((max (xn)分钟(xn)));
人物,情节(xn)
data = [xn (1,1: end-10); xn(1、2:end-9); xn (1, 3: end-8); xn (1, 4: end-7); xn(1、5:end-6); xn(1, 6:录得5个);xn (1 7: end-4); xn(1日8:end-3); xn(1日9:end-2); xn (1, 10: end-1)]; % % % % 5 giriþ1 ileri tahmin
组= xn(1、11:结束);
data =数据”;
组=组”;
指数= crossvalind (Kfold,长度(数据),3),% % % %
TestMSE_toplam = 0 (3,1);
TestMAE_toplam = 0 (3,1);
因为I = 1:3
Testy = (indices == i);
暴躁的=暴躁的;
trainy =(~暴躁的);
trainy = trainy;
TestInputData =数据(暴躁的,:)”;
TrainInputData =数据(trainy:) ';
TestOutputData =组(暴躁的,:)”;
TrainOutputData =组(trainy:) ';
Tahmin = 0(大小(TestOutputData));
Ref = 0(大小(TestOutputData));
m = 1:10
兰特(“种子”,1);
randn(“种子”,1);
网= newff(极大极小(TrainInputData (:,)), [m 1],{‘tansig’,‘tansig’,‘purelin},“trainlm”);
net.trainParam.epochs = 300;
net.trainParam.show = 20;
% net.trainparam.lr = 0.01;
net.trainparam.goal = 0.00001;
(净,tr) =火车(网,TrainInputData, TrainOutputData);
TrainOutput = sim(净,TrainInputData);
TrainMSE (m, 1) = mse (TrainOutputData-TrainOutput);
TrainMAE (m, 1) =美(TrainOutputData-TrainOutput);
TrainReport = [TrainOutput '];
TestOutput = sim(净,TestInputData);
TestMSE (m, 1) = mse (TestOutputData-TestOutput);
TestMAE (m, 1) =美(TestOutputData-TestOutput);
火车离站时刻表(m, 1) = (tr.time)之和;
结束;
(a2, b2) =找到(TestMSE = = min (min (TestMSE)));
(ax, bx) =找到(a2 = = min (a2))
a2 = a2 (ax, bx)
b2 = b2 (ax, bx)
axx(我,1)= a2;
bxx(我,1)= b2;
TestMSE_son (1) = TestMSE (a2, b2);
TestMAE_son (1) = TestMAE (a2, b2);
traintime_son(1) =火车离站时刻表(a2, b2);
%的人物,情节(TestOutputData ' O ');抓住;情节(TestOutput,‘*’);推迟
兰特(“种子”,1);
randn(“种子”,1);
网= newff(极大极小(TrainInputData (:,)), (a2(1){‘tansig’,‘tansig’,‘purelin},“trainlm”);
net.trainParam.epochs = 500;
net.trainParam.show = 20;
% net.trainparam.lr = 0.01;
net.trainparam.goal = 0.00001;
(净,tr) =火车(网,TrainInputData, TrainOutputData);
TestOutput = sim(净,TestInputData);
TestMSE_son (1) = mse (TestOutputData-TestOutput)
TestMAE_son(1) =美(TestOutputData-TestOutput)
TrainMSE_son (1) = mse (TrainOutputData-TrainOutput)
TrainMAE_son(1) =美(TrainOutputData-TrainOutput)
R2 = corrcoef (TestOutputData TestOutput);
R2 R22(1) =(1、2)^ 2
(m1, n1, r) = postreg (TestOutputData TestOutput);
注册(我,1)= r;
人物,情节(TestOutputData ' - o ');抓住;情节(TestOutput - *);推迟
Tahmin = Tahmin + TestOutput;
Ref = Ref + TestOutputData;
结束;
Tahmin_Test = Tahmin / 3;
Ref_Data = Ref / 3;
%%%%%%%%%%%%%%%%%%
TestMSE = (sum (TestMSE_son)) / 3
TestMAE = (sum (TestMAE_son)) / 3
TrainMSE = (sum (TrainMSE_son)) / 3
TrainMAE = (sum (TrainMAE_son)) / 3
Corr_R = (sum (R22)) / 3
Reg_R = ((Reg)) / 3和
人物,情节(Ref_Data);坚持;情节(Tahmin_Test ' r ')
fid1 = fopen (' hatalar_ysa.txt ', ' w + ');
流(fid1 TestMSE = % 12.6 f \ n, TestMSE);
流(fid1 TestMAE = % 12.6 f \ n, TestMAE);
流(fid1 TrainMSE = % 12.6 f \ n, TrainMSE);
流(fid1 Corr_R = % 12.6 f \ n, Corr_R);
流(fid1 Reg_R = % 12.6 f \ n, Reg_R);
米克塔= 20%米克塔表示步长
tahmin = 0 (miktar, 1);% Tahmin意味着预测
test1 = [xn(1、end-9); xn (1, end-8); xn (1, end-7); xn (1, end-6); xn(1,录得5个);xn (1, end-4); xn (1, end-3); xn (1, end-2); xn (1, end-1); xn (1)];
因为我= 1:miktar;
tahmin (1) = sim(网络,test1);
testn test1 (1, - 1) = (2, 1);
testn (2, 1) = test1 (3,1);
testn test1 (3,1) = (4,1);
testn test1 (4,1) = (1);
testn test1(5、1)= (6,1);
testn (6,1) = test1 (7, 1);
testn test1 (7 - 1) = (1);
testn test1 (8 - 1) = (9,1);
testn test1 (9,1) = (1);
testn (10 - 1) = tahmin(我,1);
test1 = testn
结束;
人物,情节(tahmin)
