Date: Mon, 22 Mar 1999 09:05:24 -0500
Sender: "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From: "Kattamuri S. Sarma" <Kattamuri.Sarma@RESPONSEINSURANCE.COM>
Subject: Re: Fitting differential equation models
Content-type: text/plain; charset=us-ascii
If you can approximate your differential equations with difference equations (it
is possible in time series
as the time is measured in years, months, or days, etc.), then you can use
SAS/ETS which can estimate individual
equations, or a system of equations. Proc Model in SAS/ETS has a number of
choices for models and estimation methods. If the observations represent
measurements at different points
of time (equidistant.... weeks, days, months etc. ) then difference equations
are discrete analogues of differential
equations. If the observations are not based on time, there should at least
be a unique way that consecutive observations
are related. The ordering of the observations is not arbitrary.
PROC MODEl is worth looking into before you start writing your code in IML.
Next to ETS, IML is the best choice
for your problem .