Data trend mining for predictive systems design

Tucker, Conrad S.

Permalink

https://hdl.handle.net/2142/24516 Copy

Description

Title

Data trend mining for predictive systems design

Author(s)

Tucker, Conrad S.

Issue Date

2011-05-25T14:27:19Z

Director of Research (if dissertation) or Advisor (if thesis)

Kim, Harrison H.M.

Doctoral Committee Chair(s)

Kim, Harrison H.M.

Committee Member(s)

• Auvil, Loretta
• Thurston, Deborah L.
• Zhang, Yuanhui

Department of Study

Industrial&Enterprise Sys Eng

Discipline

Systems & Entrepreneurial Engr

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Data Mining
• Trend Mining
• Text Mining
• Multidisciplinary Design Optimization
• Product Family
• Demand Modeling
• Multilevel Optimization
• Cyberinfrastructure
• Product Design

Abstract

The goal of this research is to propose a data mining based design framework that can be used to solve complex systems design problems in a timely and efficient manner, with the main focus being product family design problems. Traditional data acquisition techniques that have been employed in the product design community have relied primarily on customer survey data or focus group feedback as a means of integrating customer preference information into the product design process. The reliance of direct customer interaction can be costly and time consuming and may therefore limit the overall size and complexity of the customer preference data. Furthermore, since survey data typically represents stated customer preferences (customer responses for hypothetical product designs, rather than actual product purchasing decisions made), design engineers may not know the true customer preferences for specific product attributes, a challenge that could ultimately result in misguided product designs. By analyzing large scale time series consumer data, new products can be designed that anticipate emerging product preference trends in the market space. The proposed data trend mining algorithm will enable design engineers to determine how to characterize attributes based on their relevance to the overall product design. A cell phone case study is used to demonstrate product design problems involving new product concept generation and an aerodynamic particle separator case study is presented for product design problems requiring attribute relevance characterization and product family clustering. Finally, it is shown that the proposed trend mining methodology can be expanded beyond product design problems to include systems of systems design problems such as military systems simulations.

Graduation Semester

2011-05

Permalink

http://hdl.handle.net/2142/24516

Copyright and License Information

Copyright 2011 Conrad S. Tucker

Owning Collections