Developing Backward-type Hyper-portfolio Selection Framework to Evaluate Multi-layer Portfolios with Interdependent Items

• Main Authors: Liu, Ching-Fang, 劉靜方
• Other Authors: Chien, Chen-Fu
• Format: Others
• Language: en_US
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/40193238083588650919

碩士 === 國立清華大學 === 工業工程與工程管理學系 === 93 === Portfolio selection has been developed as a decision method for centuries and it is adopted in many fields for evaluating and selecting portfolios of multi-attribute item. Many studies have dealt this problem by decomposing complicated problems and focusing on single layer of portfolio to solve the subporblems. This study aims to develop a “hyper-portfolio” selection framework for the multi-layer portfolio decision problem. Multi-layer portfolio problem is defined as a combination of portfolios which are the combinations of items. In particular, Take, for example, menu consists of combination of interrelated meals which consist of combinations of interrelated food items. The proposed generic hyper-portfolio selection framework is based on backward-type selection. The vertical interactions between two layers of portfolio are derived with portfolio interactions between portfolio components. In hyper-portfolio, identified independent, interrelated and synergistic attributes perform similar properties in portfolio selection; in addition, interdependent attribute is proposed for inseparable affiliation of portfolio elements. This thesis applies the proposed framework for menu design problem with nine food item attributes during time horizon. Based on the item attributes, the portfolio selection is for meal design; furthermore, the hyper-portfolio selection evaluates menu performance by a multi-criteria mixed-integer-linear-programming (MILP) model. Finally, the thesis illustrates the hyper-portfolio MILP model calculating process, and a real nurse house menu design case. We find that hyper-portfolio selection outperforms in dealing with multi-attribute and hierarchical decision problems.