no. of samples while using 'train' for a neural network

I am solving an 'Input-Output Time-Series Problem with a Time Delay Neural Network'.

I am getting an error in the statement-

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

 error-- 'Inputs and targets have different numbers of samples'
the dimensions of the various variable are--
 inputStates = 1*2 cell 
inputs= 1*59 cell
layerStates=1*2 cell
targets=1*59 cell.

Also each column of 'inputs' cell array is a vector of 5 double values. i.e using 5 features of a problem we are predicting a target value. number of timestamps=61, input delays=2.

I dont why is the error??? inputs and targets shows the same dimensions in workspace too..

[Greg Heath]

Insufficient info. Code might help.

[Meghna]

%% Import data from spreadsheet

%% Import the data

[~, ~, raw] = xlsread('input_sheet.xlxs');

raw(cellfun(@(x) ~isempty(x) && isnumeric(x) && isnan(x),raw)) = {''};

%% Replace non-numeric cells with NaN

R = cellfun(@(x) ~isnumeric(x) && ~islogical(x),raw); % Find non-numeric cells

raw® = {NaN}; % Replace non-numeric cells

%% Create output variable

data = reshape([raw{:}],size(raw));

%% Clear temporary variables

clearvars R;

%% Import the data

[~, ~, class] = xlsread('outputsheet.xlxs');

class(cellfun(@(x) ~isempty(x) && isnumeric(x) && isnan(x),class)) = {''};

%% Replace non-numeric cells with NaN

R = cellfun(@(x) ~isnumeric(x) && ~islogical(x),class); % Find non-numeric cells

class® = {NaN}; % Replace non-numeric cells

%% Create output variable

lan=transpose(class);

z=size(data,1);

d=num2cell(data,2);

data_set=transpose(d);

inputSeries = data_set; targetSeries = lan;

% Create a Time Delay Network

inputDelays = 1:2;

hiddenLayerSize = 10;

net = timedelaynet(inputDelays,hiddenLayerSize);

% Choose Input and Output Pre/Post-Processing Functions % For a list of all processing functions type: help nnprocess

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

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

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

% Setup Division of Data for Training, Validation, Testing % For a list of all data division functions type: help nndivide

net.divideFcn = 'dividerand'; % Divide data randomly net.divideMode = 'time'; % Divide up every value net.divideParam.trainRatio = 70/100; net.divideParam.valRatio = 15/100; net.divideParam.testRatio = 15/100;

% For help on training function 'trainlm' type: help trainlm % For a list of all training functions type: help nntrain

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

% Choose a Performance Function

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

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

% Choose Plot Functions

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

net.plotFcns = {'plotperform','plottrainstate','plotresponse', ... 'ploterrcorr', 'plotinerrcorr'};

% Train the Network

[net,tr] = train(net,inputs,targets,inputStates,layerStates); %HERE IS THE ERROR

% Test the Network

outputs = net(inputs,inputStates,layerStates); errors = gsubtract(targets,outputs); performance = perform(net,targets,outputs);

% 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);

% View the Network view(net)

PLEASE REPLY AS SOON AS POSSIBLE i am new to this time series prediction.I am working on my project for m.tech thesis.

I have used Pollution Mortality example of the matlab tool box as reference.

|monospaced|

Answer 1

[Greg Heath]

I ran your code with the pollution_dataset.

The same error occurred until I placed the code after the plot function continuation symbol "..." on the next line.

I have several suggestions which are exemplified in the following code that takes advantage of defaults:

 close all, clear all, clc
[ X, T ] = pollution_dataset; % MIMO
x = cell2mat(X);
t = cell2mat(T);
whos
% Name    Size          Bytes  Class 
%   T     1x508        69088  cell                
%   X     1x508       89408  cell                
%   t     3x508       12192  double              
%   x     8x508       32512  double              
[I N ] = size(x) % [ 8 508 ]
[O N ] = size(t) % [ 3 508 ]

% Avoid overfitting: Choose H <= Hub

 Hub = -1+ ceil((0.7*N*O-O) / ( I+O+1))  % 88
ID = 0:2, H = 10     % DEFAULT 
net               = timedelaynet(ID,H);
[ Xs Xi Ti Ts ]   =  preparets(net,X,T);
ts                = cell2mat(Ts);     
MSE00s            = mean(var(ts',1)) % 102.62
net.divideFcn     = 'divideblock';
Ntrials = 10
rng('default')
for i=1:Ntrials
    net                    = configure(net,Xs,Ts);
    [ net tr Ys Es Xf Af ] = train(net,Xs,Ts,Xi,Ti);
    % NOTE: outputs = Ys and errors   = Es;
    R2(i,1)    = 1 - mse(Es)/MSE00s;
    R2trn(i,1) = 1 - mse(Es(:,tr.trainInd))/MSE00s;
    R2val(i,1) = 1 - mse(Es(:,tr.valInd))/MSE00s;
    R2tst(i,1) = 1 - mse(Es(:,tr.testInd))/MSE00s;
end
result = [(1:Ntrials)' R2 R2trn R2val R2tst ]

 % result =
% 
% 1       0.5936      0.63533      0.51213      0.47697
% 2      0.67104       0.7206      0.60579      0.50104
% 3      0.58657      0.73979      0.50454    -0.058704
% 4      0.63051      0.69725      0.60746       0.3368
% 5      0.48704      0.65898      0.34343      -0.1855
% 6    -0.016047      0.66023       0.4646      -3.7067
% 7      0.49035      0.58225      0.54201    0.0025204
% 8      0.54354      0.53697      0.58071      0.53752
% 9      0.60964      0.61695      0.56824      0.61634
% 10      0.61471      0.63631      0.63058      0.49629

 % Possible Improvements
 % 1. Check for outlier removal
 % 2. Choose ID w.r.t  input/target crosscorrelations
 % 3. Use narxnet and FD w.r.t. target autocorrelations
 % 4. Increase H  (Hub = 88)