On Thu, Dec 06, 2007 at 02:12:48PM -0600, Matthew Dennis wrote:
> I want to create an aggregate that will give the average velocity (sum of
> distance traveled / sum of elapsed time) from position and timestamps.
How do you want to handle noisy data? If you want to handle it in any
reasonable way you'd need to some sort of regression; i.e. you'd need
to consider all the relavant data and then try and minimise the total
error somehow. I've always relied on external tools to do this sort of
thing, but if you want to do it in the database you may be able to get
somewhere with pl/r.
Sam
