Performance Evaluation of Classification Algorithms in Financial Risks Prediction

Financial risks have always been the topic of interest of researchers as well as investors. Therefore, predicting financial risks in current economy is necessary. For a given dataset, selecting a suitable classifier or set of classifiers is an important task in financial risk forecast. The goal of this paper is to apply three popular machine-learning techniques; Support vector machine (SVM), Decision tree (DT) and Naïve Bayes (NB); to predicting financial risks based on real-life data - Qualitative Bankruptcy, Japanese bankruptcy and Australian credit card application. The results demonstrate that the SVM algorithm has the best and most reliable classification accuracy at 99.600%, 87.652% and 86.783% for Qualitative Bankruptcy, Japanese bankruptcy and Australian credit card application, respectively. However, the results of two algorithms (DT and NB) also yield good accuracy for three real datasets. This work also demonstrates the effectiveness of machine learning technique in classifying financial risks