Re: Modelling data with cyclical patterns?
- From: z <gzuckier@xxxxxxxxxxxxxx>
- Date: Wed, 27 Jun 2007 10:46:16 -0700
On Jun 18, 4:29 pm, "hp" <h...@xxxxxx> wrote:
Hi, I manage a death registry for a US state, and I'm fitting statistical
models to it. Old people follow a very obvious seasonal pattern, with their
deaths peaking in winter months. A linear regression model with a linear
term and a sine-wave term yielded an R-square of 0.6.
However, children follow a more complicated, pseudo-cyclical pattern. The
frequency is not necessarily seasonal -- I see overlapping multiyear
patterns.
1. Is there software in SAS or R that will automatically let you find the
phase and frequency of cyclical data?
2. Are there functions other than sine waves to consider?
Your answers will help policy planners in a life-or-death sort of way, so
thanks in advance.
SAS isn't the greatest at time-series analysis of data with cyclical
patterns; there's specialized stuff to do with that. You want stuff
like "ARIMA analysis"; here's a notice of an online course i just
randomly googled, so no recommendations or such on my part, just to
give you an idea of the subject matter. http://www.statistics.com/courses/timeseries/
price seems decent, though. I took an online grad school course in it
a few years ago, and it was tough. But at least I learned the
buzzwords.
.
- References:
- Prev by Date: Re: Variance in data sets with random trials and subjects
- Next by Date: Re: Chi square for small frequencies
- Previous by thread: Re: Modelling data with cyclical patterns?
- Next by thread: Chi square for small frequencies
- Index(es):
Relevant Pages
|
|
