Error in mapminmax.reverse

Hello everyone,

I am working on ANN code and I have to predict 365 data points ahead. I am scaling up the data between -1 and 1 using mapminmax command. The problem is that input data points are 150000 in number where as the predicted output has 365 data points. Thus I am facing a mismatch in dimension when applying reverse mapminmax command.

Data=load('Karachi_corr.csv');
length(Data);
x= Data(:,1:11);
y= Data(:,12);

[x2,PS] = mapminmax(x);                             % (150000x11)
y2 = mapminmax('apply',y,PS);                       % (150000x1)

% ANN Code 

yPred_rev = mapminmax('reverse',yPred2,PS);         % (365x1) double

Errors that it is giving me are

Error using bsxfun
Non-singleton dimensions of the two input arrays must match each other.

Error in mapminmax.reverse (line 7)
  x = bsxfun(@rdivide,x,settings.gain);

Error in nnet7.process_fcn (line 41)
      out1 = info.reverse(in2,out2);

Error in mapminmax (line 46)
    y = nnet7.process_fcn(mfilename,x,varargin{:});

Error in ANN_Func (line 130)
yPred_rev = mapminmax('reverse',yPred2,PS);

Answer 1

[Harshavardhan]

Hi @fariha Shahid

To address the dimension mismatch issue in your ANN prediction, ensure you scale your input and output separately using "mapminmax".

[x2, PSx] = mapminmax(x'); % Transpose for compatibility 

[y2, PSy] = mapminmax(y'); % Transpose for compatibility 

% ANN Code to predict yPred2

% Reverse scaling of predicted output
yPred_rev = mapminmax('reverse', yPred2, PSy);

For more information on “mapminmax” you can type the following command in a MALTAB command window

doc mapminmax

Hope this helps.