Prediction of financial crises by means of rough sets and decision trees

this paper tries to further investigate the factors behind a financial crisis. by using a large sample of countries in the period 1981 to 1999, it intends to apply two methods coming from the artificial intelligence (rough sets theory and c4.5 algorithm) and analyze the role of a set of macroeconomic and financial variables in explaining banking crises. these variables are both quantitative and qualitative. these methods do not require variables or data used to satisfy any assumptions. statistical methods traditionally employed call for the explicative variables to satisfy statistical assumptions which is quite difficult to happen. this fact complicates the analysis. we obtained good results based on the classification accuracies (80% of correctly classified countries from an independent sample), which proves the suitability of both methods.