Implementing and testing a complex interactive MOLP algorithm

摘要

Many business decisions can be modeled as multiobjective linear programming (MOLP) problems. MOLP algorithms seek solutions to these problems by interacting with decision makers to arrive at an acceptable solution. However, due in part to the increasing complexity of these algorithms, and in part to the failure of developers to use graphical user interfaces, testing and comparison of competing algorithms has been minimal. We present herein results of research designed to address this circumstance. Using widely available microcomputer tools, we designed and built a Decision Support System (DSS) capable of running MOLP algorithms, and conducted a field test which asked 98 decision makers to solve a business case using the system. Two algorithms were programmed into the DSS, one a new and more mathematically complex algorithm, and one a previously used benchmark. Results demonstrate that the more complex algorithm was preferred as a decision-making aid over the benchmark. Additionally, results show that users found the DSS equally easy to work with for both algorithms, suggesting that the graphical user interface sufficiently masked the complexity of the new algorithm. This result is encouraging for the possibility of the implementation and testing of increasingly sophisticated MOLP algorithms.