Neural Network Toolbox - Advanced AI Modeling

Neural Network Tool Box

Bashar Al asked: 2022-07-23 Deep Learning Toolbox , AI, Data Science, and Statistics , Deep Learning Toolbox , Function Approx

Design AI models with the Neural Network Toolbox. Regression, prediction, classification tools enhance machine learning projects. Start building now

Expert Answer

Kshitij Singh answered . 2025-07-09 10:10:30

% Neural Network Tool Box
% Asked by Bashar Ali on 12 Oct 2012 at 18:14
% Latest activity Edited by Image Analyst on 12 Oct 2012 at 19:30
% Hello everybody
% I am new to neural networks, but I have studied the theory and everything is
% OK. Now I've come to the practical part. I am using the Neural Network
% toolbox in Matlab, and start using NARX where x(t) is Excel file (1 column
% and 3500 rows) and y(t) is also an Excel file (1 column and 3500 rows). I
% import these files and train the system, and get the figures and everything
% is OK. My question is:
% % After doing all of the above , how do I use this network to predict the
% value of y for a given x?

I think it is as simple as

y = net(x);

% Here is the resulting script file :

% Solve an Autoregression Problem with External Input with a NARX Neural

% Network Script generated by NTSTOOL

% Created Fri Oct 12 20:48:21 IST 2012

% This script assumes these variables are defined:

% EMA - input time series.

% Close - feedback time series.

close all, clear all, clc, plt = 0; % Input purge

>inputSeries = tonndata(EMA,false,false);

>targetSeries = tonndata(Close,false,false);

Not a good name. lower case close is a MATLAB function (see above).

I'll just use X and T below

plt=plt+1;figure(plt)

plot(T)

plot(X)

% Create a Nonlinear Autoregressive Network with External Input

>inputDelays = 1:2;

Often need 0 for input (but not feedback)

>feedbackDelays = 1:2;

Use xcorr, crosscorr or a corrected version of nncorr (see below) too obtain the lags at significant peak values.

N = length(T);

ZT = zscore(T);

ZX = zscore(X);

autocorrT = nncorr(ZT,ZT,N-1);

crosscorrXTR = nncorr(ZX,ZT,N-1); % Incorrectly symmetric

crosscorrXTL = nncorr(ZT,ZX,N-1); % Incorrectly symmetric

crosscorrXT = [ crosscorrXTL(1:N-1), crosscorrXTR(N:end) ];

lags = -(N-1):N-1;

Enter code to find POSITIVE lags at significant peaks (Yeah, I could

have just used crosscorrXTR) but ..., well, you understand)

>hiddenLayerSize = 10;

Choose optimal small value ( H <= (N-1)/3 ) by trial and error.

>net = narxnet(inputDelays,feedbackDelays,hiddenLayerSize);

% Choose Input and Feedback Pre/Post-Processing Functions

% Settings for feedback input are automatically applied to feedback output

% For a list of all processing functions type: help nnprocess

% Customize input parameters at: net.inputs{i}.processParam

% Customize output parameters at: net.outputs{i}.processParam

>net.inputs{1}.processFcns = {'removeconstantrows','mapminmax'};

>net.inputs{2}.processFcns = {'removeconstantrows','mapminmax'};

Unnecessary. These are defaults

% Prepare the Data for Training and Simulation

% The function PREPARETS prepares timeseries data for a particular network,

% shifting time by the minimum amount to fill input states and layer states.

% Using PREPARETS allows you to keep your original time series data unchanged,

% while easily customizing it for networks with differing numbers of delays,

% with open loop or closed loop feedback modes.

>[inputs,inputStates,layerStates,targets] = preparets(net,inputSeries, {},targetSeries);

% Setup Division of Data for Training, Validation, Testing

% The function DIVIDERAND randomly assigns target values to training,

% validation and test sets during training.

% For a list of all data division functions type: help nndivide

>net.divideFcn = 'dividerand'; % Divide data randomly

NO. This screws up the time series uniform spacing. Use another divide

function. For example:

net.divideFcn = 'divideblock';

% The property DIVIDEMODE set to TIMESTEP means that targets are divided

% into training, validation and test sets according to timesteps.

% For a list of data division modes type: help nntype_data_division_mode

>net.divideMode = 'value'; % Divide up every value

>net.divideParam.trainRatio = 70/100;

>net.divideParam.valRatio = 15/100;

>net.divideParam.testRatio = 15/100;

Unnecessary ..

% Choose a Training Function

% For a list of all training functions type: help nntrain

% Customize training parameters at: net.trainParam

>net.trainFcn = 'trainlm'; % Levenberg-Marquardt

Unnecessary ...

% Choose a Performance Function

% For a list of all performance functions type: help nnperformance

% Customize performance parameters at: net.performParam

>net.performFcn = 'mse'; % Mean squared error

Unnecessary...

% Choose Plot Functions

% For a list of all plot functions type: help nnplot

% Customize plot parameters at: net.plotParam

>net.plotFcns = {'plotperform','plottrainstate','plotresponse', ...

'ploterrcorr', 'plotinerrcorr'};

Unnecessary ...

% Train the Network

>[net,tr] = train(net,inputs,targets,inputStates,layerStates);

tr = tr % No semicolon to expose results

% Test the Network

>outputs = net(inputs,inputStates,layerStates);

>errors = gsubtract(targets,outputs);

>performance = perform(net,targets,outputs)

Normalize and obtainR^2 statistic

NMSE = performance/var(T,1)

R2 = 1-NMSE

% Recalculate Training, Validation and Test Performance

>trainTargets = gmultiply(targets,tr.trainMask);

>valTargets = gmultiply(targets,tr.valMask);

>testTargets = gmultiply(targets,tr.testMask);

>trainPerformance = perform(net,trainTargets,outputs)

>valPerformance = perform(net,valTargets,outputs)

>testPerformance = perform(net,testTargets,outputs)

Training MSE should be adjusted to mitigate bias of using same data for training and testing.

[ I Ntrn ] = size(Xtrn)

[ O Ntrn ] = size(Ttrn)

Ntrneq = Ntrn*O % No. of training equations

ND = length( [ inputDellays, FeedbackDelays] )

Nw = (ND+1)*H +(H+1)*O % Number of estimated weights

MSEtrna = Ntrneq*MSEtrn/(Ntrneq-Nw) %Decreases with H

NMSEtrna = MSEtrna/var(Ttrn,0)

R2trna = 1-NMSEa

% View the Network

>view(net)

% Plots

% Uncomment these lines to enable various plots.

%figure, plotperform(tr)

%figure, plottrainstate(tr)

%figure, plotregression(targets,outputs)

%figure, plotresponse(targets,outputs)

%figure, ploterrcorr(errors)

%figure, plotinerrcorr(inputs,errors)

% Closed Loop Network

% Use this network to do multi-step prediction.

% The function CLOSELOOP replaces the feedback input with a direct

% connection from the outout layer.

>netc = closeloop(net);

>netc.name = [net.name ' - Closed Loop'];

>view(netc)

>[xc,xic,aic,tc] = preparets(netc,inputSeries,{},targetSeries);

>yc = netc(xc,xic,aic);

>closedLoopPerformance = perform(netc,tc,yc)

% Early 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).

ARE YOU SURE? HAVE YOU DEMONSTRATED THIS?

I THOUGHT THIS COULD ONLY HAPPENS IF THE MINIMUM FEEDBACK DELAY IS ZERO.

HOWEVER, MATLAB SHOULD THROW AN ERROR FOR A ZERO FEEDBACK DELAY

WHEN I HAVE TIME I WILL CHECK

% 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,inputSeries,{},targetSeries);

>ys = nets(xs,xis,ais);

>earlyPredictPerformance = perform(nets,ts,ys)

Neural Network Tool Box

Your academic and career questions answered by verified experts