Prediction of Closure of Accounts and Non-Performing Assets (NPA) - Regression and Attribute Oriented Induction

Abstract

This research work is fully focused on prediction technique which helps banks to predict and to avoid the specific account closure and non-performing assets. The significant of this research work has arisen owing to current shape of the banking industry further deteriorating would directly affect the economy of this country. The attribute oriented induction is used to filter the relevant attributes from noise text database. The prediction technique of multiple regression is used to highlight the factors which shall lead to the closure of accounts and non-performing assets on every stage. Further, this research work is added-up with recommended best practices for each scenario from the real time to achieve zero account closure and zero non-performing assets.

Introduction

This research focuses on predicting account closures and non-performing assets (NPAs) in the banking sector using data-driven techniques.

First Cycle – Prediction of Account Closures:

• Objective: Identify reasons for account closure and predict high-risk accounts.

• Methods:

  • Relevance Analysis & Attribute-Oriented Induction (AOI): Filters and generalizes relevant attributes from large banking datasets (complaints, transactions, marketing, risk management) to improve prediction accuracy.

  • Multiple Regression: Evaluates multidimensional factors leading to account closure, with “Customer Dissatisfaction” flagged as high-risk. The algorithm considers combined effects of issues such as unresolved complaints and transaction failures.

• Outcome: Provides timely alerts for accounts at risk of closure, allowing proactive intervention.

Second Cycle – Prediction of Non-Performing Assets (NPAs):

• Objective: Monitor asset quality to predict accounts that may become NPAs.

• Methods:

  • Linear Regression on EMI Payment Behavior: Tracks borrower repayment patterns and deviations from schedule.

  • Risk Assessment & Alerts: Calculates risk scores, triggers SMS/email reminders to borrowers, and escalates follow-ups to bank staff to prevent NPAs.

• Outcome: Enables early intervention, minimizes overdue payments, and maintains asset quality.

Conclusion

The attribute oriented relevance analysis and prediction techniques on Closure of accounts and non-performing assets shown its real worthiness in the banking industry. The era of doing this research work is thrust on the prediction techniques especially in banking industry to avoid great damage on the business and to sustain the business. The attribute oriented induction on relevance analysis deploys the strong platform to filter the very large banking database. The multiple regression method has increased the accuracy level of this prediction due to the combination of various predictor attributes. This research work shall be enhanced further on other critical areas of banking industries like prediction of money laundering, credit worthiness of the loan applicant, letter of credit, bill of lading and bank guarantee.

References

[1] Heckerman. Efficient approximations for the marginal likelihood of incomplete data given a Bayesian network. Machine Learning, 29:181–212, 2007. [2] S. Chib. Marginal likelihood from the Gibbs output. JASA, 90:1313–1321, 1995. [3] C. Carter, Markov chain Monte Carlo in conditionally Gaussian state space models. Biometrika, 83:589–601, 2006. [4] C. K. Chow and C. N. Liu. Approximating discrete probability distributions with dependence trees. IEEE Trans. on Info. Theory, 14:462–67, 1968. [5] S. Liu. Mixture Kalman filters. J. Royal Stat. Soc. B, 2000. [6] T. H. Cormen, and R. L. Rivest. An Introduction to Algorithms. MIT Press, 1990. [7] G. F. Cooper. A simple algorithm for efficiently mining observational databases for causal relationships. Data Mining and Knowledge Discovery, 2007. [8] A. Tuzhilin, “Abstract pattern discovery in databases” IEEE Trans. Knowledge and Data Engineering, 1993 [9] J. Li, “Efficient mining of emerging patterns – Discovering trends and differences”, Int. Conf. Knowledge Discovery and Data Mining,1999 [10] R. Duda and P.Hart, “Pattern Classification and Scene Analysis.”, New York, ohn Wiley & Sons, 1973. [11] D.W. Cheung, and Y. Fu, “A Fast Distributed Algorithm ,” Proc. 2006 Int’l Conf. Parallel and Distributed Information Systems, PP. 2006 Int’l Conf. Data Enf., PP. 106-114, New Orleans, Feb. 2006. [12] R. Agrawal and R. Srikant, “Mining Sequential Patterns,” Proc. 1995 Int’l Conf. Data Eng., pp.265-276,Tucson,Ariz., May 2007. [13] J.H. Friedman, “A recursive partitioning decision rule for non parametric classifiers”, IEEE Trans. On Com,1977 [14] S. Chaudhuri and U. Dayal, “An Overview of Data Warehousing and PLAP Technology,” AGM SIGMOD Record, vol. 26, pp. 65-74, 2007. [15] T. Fu and J. Han, ”Meta-Rule-Guided Mining of Association Rules in Relational Databases,” Object-Oriented Databases (KDOOD ’95), PP. 39-46, Singapore, Dec. 1995. [16] M.S. Chen, and P.S. Yu, “Data Mining: An overview from a Database Perspective,” IEEE Trans. Knowledge and Data Engg., Vol.8, PP.866-883,2006.

Copyright

Copyright © 2025 J. Stanly Thomas, Dr. N. Rajkumar. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.