PATTERN SEEKING AND SERVICE SYSTEM DESIGN DRIVEN BY FACT-BASED DECISION-MAKING

Abstract

The rapid development of Internet Technology has enabled convenient access to large amount of business transaction data and has advocated the era of "fact-based¿ decision making. An integrated ¿fact-based¿ decision-making process consists of three phases: data collecting, pattern-seeking, and performance. Motivated by the importance of understanding this process, I investigate three fundamental questions with focuses on different phases of the process. 1) Are there simple and universal patterns when people make choices? 2) How to utilize limited (insufficient) data to design a robust service system to ensure good performance under all possible situations? 3) How system design affects behavioral patterns?

This dissertation adopts a multi-theoretic and multi-disciplinary approach to offer fresh insights on the "fact-based" decision-making. I investigate three topics to tackle these questions. The first topic models a universal choice rule: the small number phenomenon. The second topic solves a robust appointment scheduling problem using a parsimonious set of information on the consultation durations. The third topic explores how information complexity and pattern recognition affect the perception of randomness.