Font Size: a A A

A model management systems approach to decisions analyzed under uncertainty: A relational database approach

Posted on:1996-06-11Degree:Ph.DType:Dissertation
University:Arizona State UniversityCandidate:Choi, ByeongkukFull Text:PDF
GTID:1468390014485775Subject:Business Administration
Abstract/Summary:
This study proposes and implements a Model Management System (MMS) for decision tree type problems in order to facilitate the encoding, the manipulation, and the solving of large-scale decision models, and to promote model sharing. The implemented MMS system combines Geoffrion's Structured Modeling Language (SML), the Information Resource Dictionary System (IRDS) as extended by Dolk, and relational database technologies.; The SML encompasses a wide variety of models in the Management Science/Operations Research (MS/OR) areas, thus providing a tool which can extend the proposed MMS into other areas in the future. The IRDS provides a way to catalogue the decision models and their components in a relational database. Model management using a relational database management system has the advantages of being able to store large decision models, flexible model management, and facilitating sharing models among multiple users.; The methodology for this research involves three phases: design, implementation, and evaluation. The design phase considers theoretical justifications, and strategies for solution algorithms. The implementation phase includes actual program coding. The evaluation phase considers how satisfactorily the research purposes are met.; In the design phase, the time and space complexities of the decision tree method, the influence diagram method, and the algebraic formulation method for decision problems are analyzed. In the implementation phase, the C language with embedded Structured Query Language (SQL) is used. The C language is used for solution algorithms, while the SQL is used for storing and fetching the required data to and from the database.; For decision tree type problems, operations including evaluation, sensitivity analysis, determination of the Value of (Im)perfect Information, transportation of the evaluation result of one decision tree to another, and modification of decision trees are implemented. Additional operations such as integrity checks for and determination of an order of node removals in an influence diagram are also implemented.; This research provides for the possibility of an integrated model management environment for Decision Analysis. To realize the full potential of this research, the system needs to be extended to include other types of decision problems, eventually leading to other MS/OR areas.
Keywords/Search Tags:Decision, Model management, System, Relational database, MMS
Related items