s-news
[Top] [All Lists]

Forecast uncertainty

To: s-news@lists.biostat.wustl.edu
Subject: Forecast uncertainty
From: yiwu ye <yiwu21111958@yahoo.com>
Date: Thu, 22 Feb 2007 23:21:32 -0800 (PST)
Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=X-YMail-OSG:Received:X-Mailer:Date:From:Subject:To:MIME-Version:Content-Type:Message-ID; b=s9IYjdi34IXc925t8D23PtmtDAC7lWif3T0QKDnlxp+quXidaMhB0lDg1Ivlm2stJdgb5hwNglMLcxABxUv53kv1TgYAjFGLbUTOinzabKb/Le4RG/yC2aMdz3ErWBJxz/6ojbSJOYZaN3z/jCxcASRcDXfwsrAOjFWVJDReIbY=;
Dear list,

This is not directly related to S+, but I think many talents in this list will be able to help. Suppose I have a very simple model, y=a+bx+error. Parameters a and b were estimated thourgh a Bayesian approach, so, I know the distributions of a, b and error. If I want to make a forecast for a given x value and qunatify its uncertainty,I think I can sample 5000 times from the distributions of a, b and error. Iinsert these smapled values into the equation gives a distribution of y, from which uncertainty can be estiamted. Am I right by doing this?

Thanks in advance,

Yiwu


Want to start your own business? Learn how on Yahoo! Small Business.
<Prev in Thread] Current Thread [Next in Thread>
  • Forecast uncertainty, yiwu ye <=