Re: How to resample continuosly time data to be represented real time ?



On 21 Mag, 18:54, "Paul E. Black" <p.bl...@xxxxxxx> wrote:
On Wednesday 21 May 2008 06:50, pamela fluente wrote:


You might try box plots
   http://en.wikipedia.org/wiki/Box_plot

As you get more values, add them to the latest box, and add a new box
when the time span (or count) is done.  When your memory is "full",
...

- Mostra testo tra virgolette -


Thank you Paul,

The representation I have to do is a simple line joining points.
Box plots are interesting but have nothing to to with the chart I
have to make.

Imagine you are receiving the prices of a stock from an imaginary
market - always open - and you
want to provide a representation of such prices by a line passing
between a (continuosly) subsampled
set of observed prices.

At each time T, i need to provide the representation of ALL the
prices observed up to T.
But clearly the chart will be drawn by joining a subset of those
points, and this subset has
maximum size 500.

I am looking for a strategy which allows me to give a good
representation, without using ALL the prices
(which could be millions if T is large) but only a subset of them
(max 500), and such subset must be
possibly recomputed each time a new price arrives.

I hope I am explaining well.

-P




.



Relevant Pages

  • Re: Slightly OT: Barcelona - Madrid HSL opens
    ... I would guess land prices are much ... cheaper and possibly easier to claim too. ... but the agency were showing us plots ... meters for prices in the range of 10,000 to 60,000 Euros. ...
    (uk.railway)
  • Re: Price2ret(seasonality in electricity prices)
    ... I am also looking for seasonality funtion especially in the electricity prices... ... I am deseasonalising by subtracting the mean untill I get QQ plots without fat tails. ... I am trying to get the sesonality funtion t use it for further use in spot market prices. ...
    (comp.soft-sys.matlab)