On Using Quantitative Methods in Decision Support Systems: Design and Implementation Issues

Abstract

The purpose of decision support systems(DSS) is to support managers, who are faced with semi- and unstructured problems. DSS aim to be integrated in the decision making process. For financial planning and modeling there exists DSS generators, however they have limited possibilities to use mathematical and statistical models. We have been involved in building several DSS’s, for location-allocation problems, for production planning, truck scheduling, personnel planning and for Box & Jenkins time series analysis which incorporate adequate mathematical tools. The first three incorporate a mixed-integer LP-package as a kernel. We have built a software library that enables us to construct and implement specific DSS’s, that incorporate an LP-package, during iterative design cycles. Our experience with very large location-allocation and GAP-type problems indicates that even for these problems solutions can be obtained, interactively, via an LP relaxation and a heuristic. This can be done since for some problems it can be proven that the number of non-integer assignments in the solution of the LP relaxation is very limited and because of the efficiency of the system. We report on recent results both with building and implementing these types of DSS and on the development of our software tools.