Date: Wed, 7 Dec 2005 07:32:46 -0800
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
Subject: Re: Forecasting Time Charter Rates
Content-Type: text/plain; charset="iso-8859-1"
The phenomena that you describe is why we have written FreeFore .
The following are some ofthe things that FreeFore checks for ..
. Is there need to transform the data to stabilize the variance of the
residuals via either power transforms ( e.g. logs,reciprocals, square
roots etc ) or is there evidented structural break points in variance .
Is there or has there been a change in parameters over time ?
suggesting that parameters estimated for the ENTIRE TIME RANGE may be
inadequate thus suggesting local estimation.
Do the residuals suggest an evidented LEVEL SHIFT , or SEASONAL PULSE
oe LOCAL TIME TREND or one or more PULSES (inliers/outliers) ?
Is there a need to incorporate lags or leads of any of your suggested
CAUSAL VARIABLES ?
For more see
If I can help ... please call
automatic forecasting systems