Modeling knowledge discovery in financial forecasting
Shang Gao, Reda Alhajj, Jon George Rokne · 2009
Knowledge discovery in financial databases has important implications. Decision making process on financial datasets is known to be difficult because of the complex knowledge domain and specific statistical characteristics of the data. In this paper, we investigate the decision making problem on financial datasets such as stock market fluctuations by means of financial ratio measurements while maintaining the interpretable results based on the association rules discovered. We approach this problem by considering different categories of financial ratios as input to the Rough Set model. A stepwise forecasting procedure is presented together with experimental results. The contribution of the paper is that we have successfully applied the static data mining techniques to the important financial domain and made a user friendly model that benefits individual investors in making investment decisions. We also discuss the extensions to embed the analysis and forecasting model into real time Enterprise Resources Planning (ERP) systems.