An interactive business intelligence platform for generic optimization based DSS : illustrated with a supply chain management problem

Abstract

The aim of this thesis is to de ne an approach to create an interactive optimization based Decision Support System (DSS) with Business Intelligence (BI) capability. The decision process involves the formulation of a problem and nding an optimal solution. The data and results associated with the model can be retrieved and analysed further to make important business decisions. We expand on this aspect of a generic framework which can be used for creating a dashboard to control the underlying model. Since decision models involve uncertainty, practical implementation of such a model includes a large number of constraints, datasets and scenarios. If the formulation of the model and its abstraction is su ciently generic, then the analysis of the data and results can be also achieved using the interactive dashboard. This dashboard is web-based and can be operated by all DSS users without any expertise in optimization system components. In terms of mathematical programming, the decision models under investigation are linear programming problems. The formulation of the models are implemented in A Mathematical Programming Language (AMPL), which is a well-established Algebraic Modeling Language (AML) for Optimization. The input and output from the modeling system are summarized into multidimensional data views for Online Analytical Processing (OLAP). The unique feature of the framework includes the instantiation of a model with di erent datasets from an interactive web-based platform. These datasets and results can be analysed and visualized further with the help of a BI tool. The generic use of the framework is presented with optimization problems from diverse domains such as Supply Chain Management and Asset Liability Management.