| The following work is the result of two years of intensive research and prototyping during the course of a graduate master's degree. In particular, this project was developed in collaboration with a group of researchers with the common goal of better integrating data mining with decision support.; Nowadays, decision makers in very competitive and diverse business sectors (i.e. management, medical, engineering, etc.) must make informed decisions about strategic problems. Such problems are often complex enough to warrant the use of sophisticated analytical methods for their resolution (i.e. forecasting, modeling, regression, etc.). Having borrowed from the myriad of techniques and models available from the areas of machine learning and statistics, data mining has become a very specialized field. Consequently, the effective application of DM is littered with many difficult and technical decisions (i.e. data cleansing, feature transformations, algorithm and parameter selection, model evaluation).; Most data mining products provide a large number of models and tools, but few provide "intelligent" assistance. For instance, few DM tool vendors provide non-specialist data miners with the ability to manage and reuse useful DM knowledge (i.e. how to perform data cleansing and feature transformations, etc.). Moreover, research seems to be based on utterly specialized DM techniques, rather than focusing on strategic, methodological, and even epistemological aspects of DM. As a result, it has been our goal to put forward a theoretical, conceptual, and technological framework for the realization of an intelligent data mining assistant, capable of empowering non-specialist data miners and decision makers. Specifically, we attempt to verify if the use of a case-based reasoning system and other artificial intelligence techniques will provide an adequate environment for the reuse of data mining knowledge with the intention of better supporting the decision making process. |
PDF Full Text Request