Re: [GENERAL] Design Question (Time Series Data)

2007-10-05 Thread Josh Tolley
On 10/4/07, Ted Byers <[EMAIL PROTECTED]> wrote: > --- Michael Glaesemann <[EMAIL PROTECTED]> wrote: > > > > > On Oct 4, 2007, at 9:30 , Ted Byers wrote: > > > > > I do not know if PostgreSQL, or any other RDBMS, > > > includes the ability to call on software such as > > "R" > > > > See PL/R: > > >

Re: [GENERAL] Design Question (Time Series Data)

2007-10-04 Thread Ted Byers
--- Michael Glaesemann <[EMAIL PROTECTED]> wrote: > > On Oct 4, 2007, at 9:30 , Ted Byers wrote: > > > I do not know if PostgreSQL, or any other RDBMS, > > includes the ability to call on software such as > "R" > > See PL/R: > > http://www.joeconway.com/plr/ > Thanks. Good to know. Ted --

Re: [GENERAL] Design Question (Time Series Data)

2007-10-04 Thread Andreas Strasser
Ted Byers wrote: If you really have such a disparity among your series, then it is a mistake to blend them into a single table. You really need to spend more time analyzing what the data means. If one data set is comprised of the daily close price of a suite of stocks or mutual funds, then it

Re: [GENERAL] Design Question (Time Series Data)

2007-10-04 Thread Michael Glaesemann
On Oct 4, 2007, at 9:30 , Ted Byers wrote: I do not know if PostgreSQL, or any other RDBMS, includes the ability to call on software such as "R" See PL/R: http://www.joeconway.com/plr/ Michael Glaesemann grzm seespotcode net ---(end of broadcast)--

Re: [GENERAL] Design Question (Time Series Data)

2007-10-04 Thread Ted Byers
--- Andreas Strasser <[EMAIL PROTECTED]> wrote: > Hello, > > i'm currently designing an application that will > retrieve economic data > (mainly time series)from different sources and > distribute it to clients. > It is supposed to manage around 20.000 different > series with differing > numb

Re: [GENERAL] Design Question (Time Series Data)

2007-10-04 Thread Andreas Strasser
Pavel Stehule schrieb: 2007/10/4, Jorge Godoy <[EMAIL PROTECTED]>: On Thursday 04 October 2007 06:20:19 Pavel Stehule wrote: I'd use the same solution that he was going to: normalized table including a timestamp (with TZ because of daylight saving times...), a column with a FK to a series table

Re: [GENERAL] Design Question (Time Series Data)

2007-10-04 Thread Pavel Stehule
2007/10/4, Jorge Godoy <[EMAIL PROTECTED]>: > On Thursday 04 October 2007 06:20:19 Pavel Stehule wrote: > > > > I did good experience with 2 variant. PostgreSQL needs 24bytes for > > head of every row, so isn't too much efective store one field to one > > row. You can simply do transformation betwe

Re: [GENERAL] Design Question (Time Series Data)

2007-10-04 Thread Jorge Godoy
On Thursday 04 October 2007 06:20:19 Pavel Stehule wrote: > > I did good experience with 2 variant. PostgreSQL needs 24bytes for > head of every row, so isn't too much efective store one field to one > row. You can simply do transformation between array and table now. But then you'll make all SQL

Re: [GENERAL] Design Question (Time Series Data)

2007-10-04 Thread Pavel Stehule
2007/10/4, Andreas Strasser <[EMAIL PROTECTED]>: > Hello, > > i'm currently designing an application that will retrieve economic data > (mainly time series)from different sources and distribute it to clients. > It is supposed to manage around 20.000 different series with differing > numbers of obse

[GENERAL] Design Question (Time Series Data)

2007-10-03 Thread Andreas Strasser
Hello, i'm currently designing an application that will retrieve economic data (mainly time series)from different sources and distribute it to clients. It is supposed to manage around 20.000 different series with differing numbers of observations (some have only a few dozen observations, other