Discover MakerZone

MATLAB and Simulink resources for Arduino, LEGO, and Raspberry Pi

Learn more

Discover what MATLAB® can do for your career.

Opportunities for recent engineering grads.

Apply Today

Thread Subject:
Neural Network - Forecast new matrix

Subject: Neural Network - Forecast new matrix

From: condor

Date: 25 Jun, 2014 17:15:09

Message: 1 of 4

Hello,

let's say I have a sequence of matrices everyone is (200 x 200).
I think that the new matrix in (t+1) can be forecasted using the previous five matrices (t, t-1, t-2, t-3, t-4).
To test this hypotheses I would like to build a neural network.

FIRST QUESTION: Should I use the Time Series Tool or the Fitting Tool?

SECOND QUESTION: If using the Fitting tool how should I write the input?
An array with 5 rows where every row contains a 200x200 matrix? or one row that contains a 1000x200 matrix?

thanks

Subject: Neural Network - Forecast new matrix

From: Greg Heath

Date: 26 Jun, 2014 03:17:13

Message: 2 of 4

"condor" wrote in message <lof02t$4k3$1@newscl01ah.mathworks.com>...
> Hello,
>
> let's say I have a sequence of matrices everyone is (200 x 200).
> I think that the new matrix in (t+1) can be forecasted using the previous five matrices (t, t-1, t-2, t-3, t-4).
> To test this hypotheses I would like to build a neural network.
>
> FIRST QUESTION: Should I use the Time Series Tool or the Fitting Tool?
>
> SECOND QUESTION: If using the Fitting tool how should I write the input?
> An array with 5 rows where every row contains a 200x200 matrix? or one row that contains a 1000x200 matrix?
>

It would be helpful to know the physical and/or mathematical significance of the matrices.

Why does it take a 200 dimensional vector to characterize whatever it is?

Does dimensionality reduction make sense?

Hope this helps

Greg

Subject: Neural Network - Forecast new matrix

From: condor

Date: 28 Jun, 2014 07:48:09

Message: 3 of 4

"Greg Heath" <heath@alumni.brown.edu> wrote in message <log3bp$1r7$1@newscl01ah.mathworks.com>...
> "condor" wrote in message <lof02t$4k3$1@newscl01ah.mathworks.com>...
> > Hello,
> >
> > let's say I have a sequence of matrices everyone is (200 x 200).
> > I think that the new matrix in (t+1) can be forecasted using the previous five matrices (t, t-1, t-2, t-3, t-4).
> > To test this hypotheses I would like to build a neural network.
> >
> > FIRST QUESTION: Should I use the Time Series Tool or the Fitting Tool?
> >
> > SECOND QUESTION: If using the Fitting tool how should I write the input?
> > An array with 5 rows where every row contains a 200x200 matrix? or one row that contains a 1000x200 matrix?
> >
>
> It would be helpful to know the physical and/or mathematical significance of the matrices.
>
> Why does it take a 200 dimensional vector to characterize whatever it is?
>
> Does dimensionality reduction make sense?
>
> Hope this helps
>
> Greg

Hi Greg,
I am doing an economic research about the CAPM. On rows I have all the observations of an economic factor in the past (1 day, 2 day, 3 day ... 200 days), on the columns I have another economic factor (1 day, 2 day, 3 day ... 200 days), the results are the returns of the company.
200 because my RAM is only 64 gb!!

And now?

Subject: Neural Network - Forecast new matrix

From: Greg Heath

Date: 28 Jun, 2014 22:11:11

Message: 4 of 4

"condor" wrote in message <lolrvp$t0i$1@newscl01ah.mathworks.com>...
> "Greg Heath" <heath@alumni.brown.edu> wrote in message <log3bp$1r7$1@newscl01ah.mathworks.com>...
> > "condor" wrote in message <lof02t$4k3$1@newscl01ah.mathworks.com>...
> > > Hello,
> > >
> > > let's say I have a sequence of matrices everyone is (200 x 200).
> > > I think that the new matrix in (t+1) can be forecasted using the previous five matrices (t, t-1, t-2, t-3, t-4).
> > > To test this hypotheses I would like to build a neural network.
> > >
> > > FIRST QUESTION: Should I use the Time Series Tool or the Fitting Tool?
> > >
> > > SECOND QUESTION: If using the Fitting tool how should I write the input?
> > > An array with 5 rows where every row contains a 200x200 matrix? or one row that contains a 1000x200 matrix?
> > >
> >
> > It would be helpful to know the physical and/or mathematical significance of the matrices.
> >
> > Why does it take a 200 dimensional vector to characterize whatever it is?
> >
> > Does dimensionality reduction make sense?
> >
> > Hope this helps
> >
> > Greg
>
> Hi Greg,
> I am doing an economic research about the CAPM. On rows I have all the observations of an economic factor in the past (1 day, 2 day, 3 day ... 200 days), on the columns I have another economic factor (1 day, 2 day, 3 day ... 200 days), the results are the returns of the company.
> 200 because my RAM is only 64 gb!!
>
> And now?

What is CAPM?

Your description doesn't make sense to me. Input and output examples are column vector pairs. "I"nputs are I-dimensional and "o"utputs are O-dimensional. So, for N pairs of examples:

 [ I N ] = size(input)

 [ O N ] = size(target)

Can you describe your problem w.r.t. this point of view?

Hope this helps.

Greg

Tags for this Thread

What are tags?

A tag is like a keyword or category label associated with each thread. Tags make it easier for you to find threads of interest.

Anyone can tag a thread. Tags are public and visible to everyone.

Contact us