|
Activity Number:
|
279
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistics and Marketing
|
| Abstract - #310328 |
|
Title:
|
Statistical Issues in Optimal Product Design
|
|
Author(s):
|
James Cochran*+
|
|
Companies:
|
Louisiana Tech University
|
|
Address:
|
College of Business, Ruston, LA, 71272,
|
|
Keywords:
|
Resampling ; Marketing ; Bias ; Optimization ; Political Science
|
|
Abstract:
|
Recent advances in algorithms and formulations enable us to solve product design problems to demonstrable optimality in a few seconds. Since sample data are used to estimate parameter values for these problems, the problems lie at the juncture of statistics and operations research; the maximands of these problems are estimates with associated sampling distributions and statistical characteristics. Our ability to rapidly solve these problems to optimality enables us to use resampling approaches that were heretofore computationally impractical to assess various statistical characteristics. We provide an overview of some product design problems that can now quickly be solved to optimality (including the single product design and political platform problems) and present approaches for assessing the variation and bias of sampling distributions associated with the maximands of such problems.
|
