iterative algorithm for calculating the moving average
- From: "me" <me@xxxxxx>
- Date: Fri, 29 Apr 2005 02:53:17 GMT
Hi group, I have an interesting problem.
I would like to calculate the moving average of a time series at certain
intervals (for some variables I need a 1 hr average, and for others a 4 hr
average). I need this for quality control purposes at the plant I work. I
tried using the "standard" method (i.e. keep all the values in an array and
divide by the horizon), but this requires too much resources from the online
system (even when I use a "revolving" array and I have no FOR..NEXT loops).
I also tried approximating the moving average with a discrete exponential
filter of the form
average(t)=lambda*average(t-1) +(1-lambda)*x(t)
this last algorithm is fast, but when the signal is noisy, it does not work
very well. Also, it is always a lot of work to find the lambda that
approximates the moving average I want.
Any ideas? Does anyone have any algorithms or references on the above
subject?
Thanks!
Yannis
.
- Follow-Ups:
- Re: iterative algorithm for calculating the moving average
- From: glenbarnett
- Re: iterative algorithm for calculating the moving average
- From: Ron Shepard
- Re: iterative algorithm for calculating the moving average
- Prev by Date: Re: stop-test for iterative methods
- Next by Date: Re: iterative algorithm for calculating the moving average
- Previous by thread: Chebyshev interpolation polynomial.
- Next by thread: Re: iterative algorithm for calculating the moving average
- Index(es):
Relevant Pages
|