Equation solving and Regress
- To: mathgroup at smc.vnet.net
- Subject: [mg30985] Equation solving and Regress
- From: "Mark S. Coleman" <mcoleman at bondspace.com>
- Date: Sat, 29 Sep 2001 04:19:04 -0400 (EDT)
- Sender: owner-wri-mathgroup at wolfram.com
Hi, This is a something of a follow-up to another question I posted on econometric models a few days ago. I'm using the Regress command to estimate a series of equations that involved lagged dependent variables, e.g., y(t) = a + b*y(t-1) + c*x(t) + d*z(t), etc., and then extracted the fitted equation using the RegressionReport->BestFit option. Since the fitted equation is a difference equation, it is not immediately obvious to me how I could evaluate the resulting rule to find values of y(t) (assuming we know x and z). I know that I can re-write this rule and use FoldList, but I'm wondering if there is a way to evaluate the BestFit rule directly, in a way that correctly reflects the recurrence. Thanks, Mark