Precisions obtained with network openloop NarX

1 次查看（过去 30 天）
FRANCISCO
FRANCISCO 2013-3-3
Write data from network openloop with NarX:
N=2200
Ndof=1449
Neq=2200
Ntrneq=1540
Nw=91
I=8
O=1
R2=0.9911
R2a=0.9907
R2trn=0.9914
R2tst=0.9900
R2val=0.9905
MSE=8.48e-5
MSEa=8.84e-5
MSEtrn=9.2e-5
MSEtst=9.2e-5
MSEval=9.065e-5
If you know someone tell me if there is any sign of overfitting appreciate it because I think it achieves high precision and the graph I get a little offset from the original. Variables used are highly correlated. We also appreciate and would tell me ontener several steps ahead with real data and because I get to perform one iteration morning values but need to continue iterating my future values of inputs and that is not possible. thank you very much

请先登录，再回答此问题。

采纳的回答

Greg Heath
Greg Heath 2013-3-4
net = narxnet(ID,FD,H);
What are ID, FD and H?
Typically, the sign of overfitting is very high R2trn with much lower R2trna, R2val and R2tst.
You're looking good.
Greg
  1 个评论
FRANCISCO
FRANCISCO 2013-3-4
ID = 1:2
FD = 1:2
H = I think more or less I move between 8-12. Is that I have made many tests and now I'm in college and I have my data here, but the number of H is handled in that interval.
As calculated R2trna? It is the only parameter that would fail me by check. All other conditions met.
Many thanks Greg

请先登录，再进行评论。

更多回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Deep Learning Toolbox 的更多信息

标签

产品

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by