i get an error when creating newff please help me ..

the code is attached

the excel file has the dataset. it has 24 columns. first 12 columns has the target value 1. the last 12 colums have the target value 0. i wanted to design a NN for classification into 1 or 0. the data was copied to dat file. i selected all the 198 rowsand 24 colums to input.dat . single rown with 24 colums as target.dat and i get the following error . please rectify.

Error:

Error using newff (line 107)
Input ranges is not a two column matrix.
Error in signature1 (line 17)
net=newff(input_n,target_n,13);

[Geoff Hayes]

Is newff a function of the Neural Network Toolbox? If so, which version of MATLAB are you using? And what can you say about the inputs input_n and target_n - what are their dimensions?

Answer 1

[Greg Heath]

It looks like you are using an obsolete version of NEWFF where the first input is minmax(input_n) instead of input_n.

However, even the latest version of NEWFF is obsolete.

What version of MATLAB do you have? Type ver into the command line

The current versions of MLP are

 Regression/curve-fitting: FITNET(calls FEEDFORWARDNET) instead of NEWFIT(calls NEWFF)

 Classification/pattern-recognition: PATTERNNET(calls FEEDFORWARDNET)instead of NEWPR(calls FEEDFORWARDNET)

Hope this helps.

Thank you for formally accepting my answer

Greg

[Dhandapani.S]

8.0.0.783 (R2012b) is the version of matlab installed

[Dhandapani.S]

error = Error using newff (line 107) Input ranges is not a two column matrix. Error in nn1 (line 8) net = newff(inputs,targets,H)

code clc; clear all; load wedinput.dat; load wedtarget.dat; inputs=wedinput(1:24,1:198); targets=wedtarget(1:24,1); H = 10; net = newff(inputs,targets,H); net.divideParam.trainRatio =.5; net.divideParam.testRatio =.25; net.divideParam.valRatio =.25; net.plotFcns = {'ploterrhist'};

matlab version is 8.0.0.783 (R2012b) is the version of matlab installed. please help to solve the issue

[Greg Heath]

Reread my answer about minmax(inputs)

[xingxingcui]

matlabR2010 above use feedforwardnet function instead.