GRNN gives different results when input dataare normalized to [0,1] ?

Discussion in 'MATLAB' started by Eric Parent, May 24, 2005.

  1. Eric Parent

    Eric Parent Guest

    Hi,

    I am using a GRNN to approximate data and it works fine. I tried
    to normalize (linearly) my data on the [0,1] set and rerun the
    simulation on a new GRNN but the output data is not good at all. I
    looked at my normalisation method and it seems OK.
    Is there any particular reason why it's like that ? Does it has to
    do with the spread ? I left it as is, thinking it was self-adjusted
    by the newgrnn() function.

    Any hints are wellcome.

    Eric
     
    Eric Parent, May 24, 2005
    #1
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.