Knowledge mining methodologies for economic prediction

Abstract

The objective of knowledge discovery and data mining lies in the generation of useful insights from a store of data. This thesis presents a framework for knowledge mining to provide a systematic approach to the prediction of economic variables for strategic investment and marketing. Isolated approaches using statistical methods or artificial intelligence techniques have their limitations. These limitations underscore the need for a general framework to support a multistrategy approach to economic prediction. Previous studies of economic growth have largely focused on the manual construction of casual models using linear techniques. For instance, the prediction of economic variables such as gross domestic product (GDP) have largely focused on linear models such as multivariate regression. However, such models depart from reality due to the prevalence of nonlinear factors in economic systems. Among nonlinear models, neural networks have been applied to the task of forecasting. Another pertinent methodology for nonlinear prediction is case based reasoning (CBR). An important advantage of case based reasoning is its ability to deal with small data sets. For this reason, it can be applied to many domains with limited observations, in contrast to the large data sets required by many other techniques such as neural networks. These ideas are investigated in a practical domain: predicting economic performance in emerging markets in order to support decision making for direct foreign investment.