"Andrew Bullock" <(E-Mail Removed)> wrote in message

news:Yi9Nd.704$(E-Mail Removed)...

> Hi guys i hope you can help me out with this one,

>

> Im trying to construct a very simple method of time series predicition

> (i think thats what its called, ill explain anyway incase im wrong).

>

>

> I have an arbitrary window of time every minute in which i can perform

> an operation. I want to perform this operation as close to the centre of

> the window as possible (dont ask why, this is a very basic, alsmost

> analogous description of the problem). I know the start time and length

> of the past, say.. 20 time slots. How can i predict the middle of the

> next window?

>

> The windows arent randomly placed, but i cant say exactly how they are

> placed :S

> All i can say about them is that any one window will overlap the

> previous window by some amount, usually more than less.

>

> I've been looking at FFTs and regression but im a bit stuck as to how

> get something simple implemented.

>

> Any help anyone could give me would be greatly appreciated. Thanks a lot

> in advance.
This isn't a question about C++, but about math, so it's

not topical here. I'll wager the good folks over at the

'sci.math' group could help you out. Once you've devised

your formula, if you need help expressing it in C++,

post back here, and we'll try to help.

Purpose of comp.lang.c++:

http://www.slack.net/~shiva/welcome.txt
-Mike