Improving Financial Bankruptcy Prediction Using Oversampling Followed by Fuzzy Rough Feature Selection via Evolutionary Search

Abstract

Recently, bankruptcy prediction has been addressed as one of the most interesting as well as challenging issues for financial institutions and business. For creditors and investors, bankruptcy prediction plays very interesting role by facilitating decision-making ability in various areas such as business, accounting, and finance etc. Due to assemblage of large volume of inconsistent, highly imbalanced, irrelevant and redundant data from companies and other creditors, its always a very challenging and complex task to handle financial risk associated with company by developing an effective prediction model. This chapter presents a new methodology for improving the bankruptcy prediction performance of various machine learning algorithms. Firstly, we convert imbalanced dataset consisting of bankrupt and non-bankrupt into balanced dataset by applying oversampling technique. Then, relevant and non-redundant features are generated based on fuzzy rough feature selection technique via evolutionary search. Furthermore, performance of various machine learning algorithms are fully analysed by applying them on this highly balanced reduced dataset. Moreover, discriminating ability of different features are analysed based on feature ranking algorithm. Finally, we present a comparative study of our best results with already existing results. From different experimental results, we can conclude that bankruptcy prediction can be enhanced by suitably adjusting the class distribution followed by fuzzy rough set based feature selection via evolutionary search. © 2021, The Author(s), under exclusive license to Springer Nature Switzerland AG.