Mariana Valério de Carvalho, investigadora do grupo IST, finaliza doutoramento
Tema da Tese: Enhancing the Process of View Selection in Data Cubes using What-If analysis
Autor: Mariana Reimão Queiroga Valério de Carvalho
Programa Doutoral: Programa Doutoral em Informática
Orientador: Orlando Manuel Oliveira Belo
Data de Defesa: 04/09/2019
To compete in today’s society, enterprise managers need to be able to deal with the arising challenges of the competitive market. The increasing competition and the amount of electronic information imply new challenges related to decision-making processes. Collecting relevant information and using Business Intelligence tools are determining factors in decision-making processes and in gaining competitive advantage.
However, gathering and storing relevant information may not be enough. The possibility of simulating business hypothetical scenarios could be the advantage that companies need. What-If analysis can help to achieve this competitive advantage.
What-If analysis allows to create simulation models to explore the behaviour of a system, by analyzing the effects of changing values of parameters, which cannot otherwise be discovered by a manual analysis of historical data, and so, allowing the analysis of the consequences of those changes.
A successful What-If analysis process depends mainly on the user experience, his/her knowledge about the business information and the What-If analysis process itself. Otherwise, it can turn into a long and difficult process, especially in the choice of input parameters for the analysis.
In this doctoral thesis, a hybridization methodology is proposed that integrates OLAP preferences in the conventional process of What-If analysis. This integration aims to discover the best recommendations for the choice of input parameters for the analysis scenarios using OLAP preferences, helping the user to overcome the difficulties that normally arise in conventional What-If analysis process. The developed methodology helps to discover more specific, oriented and detailed information that could not be discovered using the conventional What-If analysis process.