Search All of the Math Forum:

Views expressed in these public forums are not endorsed by NCTM or The Math Forum.

Notice: We are no longer accepting new posts, but the forums will continue to be readable.

Topic: Improving ANN results
Replies: 16   Last Post: Nov 13, 2013 9:10 AM

 Messages: [ Previous | Next ]
 Greg Heath Posts: 6,387 Registered: 12/7/04
Re: Improving ANN results
Posted: Nov 6, 2013 1:16 AM

"Greg Heath" <heath@alumni.brown.edu> wrote in message <l5cm8f\$ak9\$1@newscl01ah.mathworks.com>...
> "Greg Heath" <heath@alumni.brown.edu> wrote in message <l56r3v\$1c9\$1@newscl01ah.mathworks.com>...
> > "chaudhry " <bilal_zafar9@yahoo.com> wrote in message <l55259\$jmo\$1@newscl01ah.mathworks.com>...
> > > "Greg Heath" <heath@alumni.brown.edu> wrote in message <l4f7fl\$dv5\$1@newscl01ah.mathworks.com>...
> > > > "chaudhry " <bilal_zafar9@yahoo.com> wrote in message <l4a24h\$33\$1@newscl01ah.mathworks.com>...
> > > GREG
> > >
> > > chk ur mail....i have send u my excel database.....
> > >
> > > and i have snt u my code also..
> > > so see whats the problem in it and how it can solved
> > >
> > > because my mse value coming is very large
> > >
> > > give me solution to my problem kindly as early as possible

> >
> > Unable to open. Send *.txt or *.m

>
> Thanks.
>
> 1. Using minmax I found that x1 is constant. Why in the world would you send me data with a constant input???
>
> 2. Removing the constant input, the correlation coefficient matrix for [x ; t ]' indicates
> that only x2 is significantly linearly correlated with t.
>
> % minmaxxt = 8600 11666
> % 76 105000
> % 841 334960
> %
> % CC = 1.0000 -0.0612 -0.1399
> % -0.0612 1.0000 0.7929
> % -0.1399 0.7929 1.0000
> %
> % P = 1.0000 0.5924 0.2189
> % 0.5924 1.0000 0.0000
> % 0.2189 0.0000 1.0000
> %
> % SIGMASK = 0 0 0
> % 0 0 1 ==> only x2 and t
> % 0 1 0
>
> 3. Plotting x1 , x2, t, t vs x1 and t vs x2 indicate that there is very little hope of getting
> a good nonlinear model from this data unless outliers are removed.
>
> 4. Standardize the data using zscore or mapstd. Remove outliers and start over.

I used the term 'outlier' because those points are nothing like the others. They may be valid measurements. However, they cannot contribute to a sensible model unless the gap is filled with more data.

Hope this helps.

Greg

Date Subject Author
10/13/13 chaudhry
10/15/13 Greg Heath
10/15/13 Greg Heath
10/17/13 chaudhry
10/17/13 chaudhry
10/18/13 Greg Heath
10/17/13 chaudhry
10/17/13 chaudhry
10/23/13 chaudhry
10/23/13 chaudhry
10/23/13 chaudhry
10/25/13 Greg Heath
11/3/13 chaudhry
11/3/13 Greg Heath
11/6/13 Greg Heath
11/6/13 Greg Heath
11/13/13 chaudhry