Đang chuẩn bị liên kết để tải về tài liệu:
Predicting bankruptcy using machine learning algorithms

Không đóng trình duyệt đến khi xuất hiện nút TẢI XUỐNG Tải xuống

Bankruptcy prediction is of great utility for all economic stakeholders. Therefore, diverse methods have been applied for the early detection of financial risks in recent years. The objective of this paper is to propose an ensemble artificial intelligence (AI) model for effectively predicting the bankruptcy of a company. | 6 Thi Kha Nguyen, Thi Phuong Trang Pham PREDICTING BANKRUPTCY USING MACHINE LEARNING ALGORITHMS Thi Kha Nguyen1, Thi Phuong Trang Pham2 The University of Danang - Campus in Kontum; nguyenkha130490@gmail.com 2 The University of Danang - University of Technology and Education; ptptrang@ute.udn.vn 1 Abstract - Bankruptcy prediction is of great utility for all economic stakeholders. Therefore, diverse methods have been applied for the early detection of financial risks in recent years. The objective of this paper is to propose an ensemble artificial intelligence (AI) model for effectively predicting the bankruptcy of a company. This study is designed to assess various classification algorithms over two bankruptcy datasets - Polish companies bankruptcy and Qualitative bankruptcy. The comparison results show that the bagging-ensemble model outperforms the others in predicting bankruptcy datasets. In particular, with the test data of Polish companies bankruptcy, the regression tree learner bagging (REPTree-bagging) ensemble model yields an accuracy of 100%. In predicting Qualitative bankruptcy dataset, the Random tree bagging (RTree-bagging) ensemble model has the highest accuracy with 96.2% compared to other models. Key words - Bankruptcy prediction; single-methods; ensemblemodels; artificial intelligence methods; bagging. 1. Introduction Financial risk prediction is one of a critical topic in the domain of financial analysis because it can help companies to reduce financial distress and take appropriate actions in the future. Many financial risk prediction tasks are basically binary classification problems, which means observations are assigned to one of the two groups after data analysis [1]. This paper focuses on classifying bankruptcy problems. Thanks to the development of computer power and data storage technologies, classification algorithms can be used to quickly and effectively predict financial data. However, the algorithm evaluation or algorithm selection .