hello how can i use neural network to model my rainfall data?

 what does training and validation and all mean in it do i need to calculate these things or are they given by matlab itself?

Answer 1

[Greg Heath]

NN regressors/curve-fitters are based on N examples of pairs of I-dimensional input data vectors and the corresponding O-dimensional target output vectors.

So to begin, create your input and target matrices.

[ I N ] = size(input)

[ O N ] = size(target)

Then see the documentation and examples for fitnet.

doc fitnet

help fitnet

Finally check NEWSGROUP and ANSWERS for posts using the searchword

fitnet.

Hope this helps.

Greg

[kushbou]

ok thanks...but i wanted to know how do we actually decide upon the percentage for the training,tests,number of hidden neurons etc...do we have to put it there or the toolbox itself determines the number of neurons etc

[Greg Heath]

That is why I recommended reading the documentation and studying the examples.

Then you can practice with one of the MATLAB data sets with which we are familiar

help nndatasets

Greg