Predict one independent variable (IV) from 25 dependent variables (DV)

Hi,

Besides using the Neural Network toolbox, is there a way to - for lack of a better explanation - let Matlab create a function that can predict the indepdendent variable? I have 167 data points of one IV and 25 DV.

I have previously used an empirical function and unconstrained non-linear optimization, but was wondering if there is a smarter way than just basically guessing a function that could fit. Note that I do not need to use all of the DVs, but could start with 3 or 4 DVs.

Many thanks in advance! Jan

Answer 1

[Greg Heath]

Even if you do not want to use a net, I doubt if you need all 25 to model 1.

I have successfully used the following strategy:

 1. 25 plots
 2. A sequence of trimmed polynomial models
   (TRIMMED ==> negligible terms are omitted)

   a. Linear
   b. Quadratic
   c. I have never had to use higher powers

Hope this helps.

Thank you for formally accepting my answer

Greg

[Jan Norbisrath]

Hei Greg! How do you trim a polynomial? I could not find much by googling. Thanks! Jan