Prev: Locating GUIDE Gui Figure Handle/Passing Data
Next: image transfer using instrument control toolbox
From: Luna Moon on 31 May 2010 21:39
Hi all,

Here are the data I am working on:

dDataTmp2 is a Nx2 matrix, with the first column being the timestamps
and the second column the data.

K>> datestr(dDataTmp2(1, 1))

ans =

01-Apr-2010 00:19:00

I would like to compute the average of the second-column-data every M
seconds, e.g. M=5, and put the results into a new smaller matrix.

Of course I can do datestr on each timestamp and match the "seconds",
but that's slow and I have to do this repetitively and iteratively on
GB of data.

So the faster approach is to directly operate on the date/time
numbers; but I don't know how to operate on them.

Anybody could think of some "the fastest" algorithms for this?

Thank you!

From: TideMan on 31 May 2010 22:29
On Jun 1, 1:39 pm, Luna Moon <lunamoonm...(a)gmail.com> wrote:
> Hi all,
>
> Here are the data I am working on:
>
> dDataTmp2 is a  Nx2 matrix, with the first column being the timestamps
> and the second column the data.
>
> K>> datestr(dDataTmp2(1, 1))
>
> ans =
>
> 01-Apr-2010 00:19:00
>
> I would like to compute the average of the second-column-data every M
> seconds, e.g. M=5, and put the results into a new smaller matrix.
>
> Of course I can do datestr on each timestamp and match the "seconds",
> but that's slow and I have to do this repetitively and iteratively on
> GB of data.
>
> So the faster approach is to directly operate on the date/time
> numbers; but I don't know how to operate on them.
>
> Anybody could think of some "the fastest" algorithms for this?
>
> Thank you!

You omitted the most vital piece of information:
are the data equispaced in time?
From: John D'Errico on 31 May 2010 23:18
Luna Moon <lunamoonmoon(a)gmail.com> wrote in message <357ca75a-8456-4100-9362-4e6c775cfa11(a)e21g2000vbl.googlegroups.com>...
> Hi all,
>
> Here are the data I am working on:
>
> dDataTmp2 is a Nx2 matrix, with the first column being the timestamps
> and the second column the data.
>
> K>> datestr(dDataTmp2(1, 1))
>
> ans =
>
> 01-Apr-2010 00:19:00
>
> I would like to compute the average of the second-column-data every M
> seconds, e.g. M=5, and put the results into a new smaller matrix.
>
> Of course I can do datestr on each timestamp and match the "seconds",
> but that's slow and I have to do this repetitively and iteratively on
> GB of data.
>
> So the faster approach is to directly operate on the date/time
> numbers; but I don't know how to operate on them.
>
> Anybody could think of some "the fastest" algorithms for this?
>
> Thank you!

Use my consolidator function, keying on the
indicator:

floor(timestamps*24*60*12)

for your averaging. This allows you to not worry
if the data is equally or unequally spaced.

Find consolidator on the FEX:

http://www.mathworks.com/matlabcentral/fileexchange/8354

HTH,
John
From: Luna Moon on 1 Jun 2010 08:19
On May 31, 10:29 pm, TideMan <mul...(a)gmail.com> wrote:
> On Jun 1, 1:39 pm, Luna Moon <lunamoonm...(a)gmail.com> wrote:
>
>
>
> > Hi all,
>
> > Here are the data I am working on:
>
> > dDataTmp2 is a  Nx2 matrix, with the first column being the timestamps
> > and the second column the data.
>
> > K>> datestr(dDataTmp2(1, 1))
>
> > ans =
>
> > 01-Apr-2010 00:19:00
>
> > I would like to compute the average of the second-column-data every M
> > seconds, e.g. M=5, and put the results into a new smaller matrix.
>
> > Of course I can do datestr on each timestamp and match the "seconds",
> > but that's slow and I have to do this repetitively and iteratively on
> > GB of data.
>
> > So the faster approach is to directly operate on the date/time
> > numbers; but I don't know how to operate on them.
>
> > Anybody could think of some "the fastest" algorithms for this?
>
> > Thank you!
>
> You omitted the most vital piece of information:
> are the data equispaced in time?

Hi Tideman,

No, the data are not equally spaced in time. Any more thoughts?

Thanks a lot!
From: Luna Moon on 1 Jun 2010 08:34
On May 31, 11:18 pm, "John D'Errico" <woodch...(a)rochester.rr.com>
wrote:
> Luna Moon <lunamoonm...(a)gmail.com> wrote in message <357ca75a-8456-4100-9362-4e6c775cf...(a)e21g2000vbl.googlegroups.com>...
> > Hi all,
>
> > Here are the data I am working on:
>
> > dDataTmp2 is a  Nx2 matrix, with the first column being the timestamps
> > and the second column the data.
>
> > K>> datestr(dDataTmp2(1, 1))
>
> > ans =
>
> > 01-Apr-2010 00:19:00
>
> > I would like to compute the average of the second-column-data every M
> > seconds, e.g. M=5, and put the results into a new smaller matrix.
>
> > Of course I can do datestr on each timestamp and match the "seconds",
> > but that's slow and I have to do this repetitively and iteratively on
> > GB of data.
>
> > So the faster approach is to directly operate on the date/time
> > numbers; but I don't know how to operate on them.
>
> > Anybody could think of some "the fastest" algorithms for this?
>
> > Thank you!
>
> Use my consolidator function, keying on the
> indicator:
>
>     floor(timestamps*24*60*12)
>
> for your averaging. This allows you to not worry
> if the data is equally or unequally spaced.
>
> Find consolidator on the FEX:
>
> http://www.mathworks.com/matlabcentral/fileexchange/8354
>
> HTH,
> John

Hi John,

Thanks a lot! But after reading the help of Consolidator, I just
couldn't figure out how to use it in the right way to solve my
problem.

Could you please elaborate a bit further?

Thank you!
 |  Next  |  Last
Pages: 1 2
Prev: Locating GUIDE Gui Figure Handle/Passing Data
Next: image transfer using instrument control toolbox