How forecast One step ahead (N+1) with NARNET?
2 views (last 30 days)
Show older comments
Hello people! Please, can you help me?
I want predict One Step beyond original data using NARNET. Original data has 62000 steps. Want know the 62001.
But what I get for prediction is the exact same existing steps in training data. (Same N steps, same curves)
Why network can't predict next future step beyond original data end? I used Removedelay. (Want the N+1)
Attached are: Original data (EURUSD), performance stats. Bellow, the code used.
Thanks, and God bless you all! Eric
% Solve an Autoregression Time-Series Problem with a NAR Neural Network
% Script generated by Neural Time Series app
% This script assumes this variable is defined:
% EURUSD - feedback time series.
T = tonndata(EURUSD,true,false);
% Choose a Training Function
trainFcn = 'trainbr'; % Bayesian Regularization backpropagation.
% Create a Nonlinear Autoregressive Network
feedbackDelays = 1:1;
hiddenLayerSize = 30;
net = narnet(feedbackDelays,hiddenLayerSize,'open',trainFcn);
% Choose Feedback Pre/Post-Processing Functions
net.input.processFcns = {'removeconstantrows','mapminmax'};
net.trainParam.min_grad = 3e-8;
% Prepare the Data for Training and Simulation
[x,xi,ai,t] = preparets(net,{},{},T);
% Setup Division of Data for Training, Validation, Testing
net.divideFcn = 'dividetrain';
net.divideMode = 'time'; % Divide up every sample
% Choose a Performance Function
net.performFcn = 'mse'; % Mean Squared Error
% Choose Plot Functions
net.plotFcns = {'plotperform','plottrainstate', 'ploterrhist', ...
'plotregression', 'plotresponse', 'ploterrcorr', 'plotinerrcorr'};
% Train the Network
[net,tr] = train(net,x,t,xi,ai);
% Test the Network
y = net(x,xi,ai);
e = gsubtract(t,y);
performance = perform(net,t,y)
% Step-Ahead Prediction Network
% For some applications it helps to get the prediction a timestep early.
% The original network returns predicted y(t+1) at the same time it is
% given y(t+1). For some applications such as decision making, it would
% help to have predicted y(t+1) once y(t) is available, but before the
% actual y(t+1) occurs. The network can be made to return its output a
% timestep early by removing one delay so that its minimal tap delay is now
% 0 instead of 1. The new network returns the same outputs as the original
% network, but outputs are shifted left one timestep.
nets = removedelay(net);
nets.name = [net.name ' - Predict One Step Ahead'];
view(nets)
[xs,xis,ais,ts] = preparets(nets,{},{},T);
ys = nets(xs,xis,ais);
stepAheadPerformance = perform(nets,ts,ys)
0 Comments
Accepted Answer
Greg Heath
on 21 Mar 2016
Do not use the REMOVEDELAY command
It is not necessary
and
it is too confusing.
If you need detailed help, use one of the MATLAB example sets
help nndatasets
and/or
doc nndatasets.
Almost anything you need to do is in a former post. Try searching in BOTH the NEWSGROUP and ANSWERS using
greg narnet
Check the most recent posts first
Hope this helps
Thank you for formally accepting my answer
Greg
2 Comments
Greg Heath
on 23 Mar 2016
More clear if you use "o" for openloop (OL) instead of "s".
After designing a successful OL configuration, close the loop to predict beyond the known target region.
Reread the help and doc documentation re converting OL designs to CL designs.
Also search in BOTH NEWSGROUP and ANSWERS for my posts
greg narnet
Read the most recent posts first.
Greg
More Answers (0)
See Also
Categories
Find more on Sequence and Numeric Feature Data Workflows in Help Center and File Exchange
Products
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!