Machine Learning Techniques in Bank Credit Analysis (2024)

Authors: Fernanda M. Assef, Maria Teresinha A. Steiner

Abstract:

The aim of this paper is to compare and discuss better classifier algorithm options for credit risk assessment by applying different Machine Learning techniques. Using records from a Brazilian financial institution, this study uses a database of 5,432 companies that are clients of the bank, where 2,600 clients are classified as non-defaulters, 1,551 are classified as defaulters and 1,281 are temporarily defaulters, meaning that the clients are overdue on their payments for up 180 days. For each case, a total of 15 attributes was considered for a one-against-all assessment using four different techniques: Artificial Neural Networks Multilayer Perceptron (ANN-MLP), Artificial Neural Networks Radial Basis Functions (ANN-RBF), Logistic Regression (LR) and finally Support Vector Machines (SVM). For each method, different parameters were analyzed in order to obtain different results when the best of each technique was compared. Initially the data were coded in thermometer code (numerical attributes) or dummy coding (for nominal attributes). The methods were then evaluated for each parameter and the best result of each technique was compared in terms of accuracy, false positives, false negatives, true positives and true negatives. This comparison showed that the best method, in terms of accuracy, was ANN-RBF (79.20% for non-defaulter classification, 97.74% for defaulters and 75.37% for the temporarily defaulter classification). However, the best accuracy does not always represent the best technique. For instance, on the classification of temporarily defaulters, this technique, in terms of false positives, was surpassed by SVM, which had the lowest rate (0.07%) of false positive classifications. All these intrinsic details are discussed considering the results found, and an overview of what was presented is shown in the conclusion of this study.

Keywords: Artificial Neural Networks, ANNs, classifier algorithms, credit risk assessment, logistic regression, machine learning, support vector machines.

ProcediaAPABibTeXChicagoEndNoteHarvardJSONMLARISXMLISO 690PDF Downloads 1165

References:

[1]Fenerich, A. T., Steiner, M. T. A., Steiner Neto, P. J., Tochetto, E. A., Tsutsumi, D. P., Assef, F. M., dos Santos, B. S. 2018. Use of Machine Learning techniques in bank risk analysis. Proceedings of International Conference on Computers and Industrial Engineering (CIE48), Auckland, New Zealand, 2-5 December.
[2]Fahner, G. 2012. Estimating causal effects of credit decisions, International Journal of Forecasting, 28(1), pp. 248–260.
[3]Zhang Z., Gao G., Shi Y. 2014. Credit risk evaluation using multi-criteria optimization classifier with kernel, fuzzification and penalty factors. European Journal of Operational Research. 237(1), pp. 335-348.
[4]Luo, S., Kong, X., Nie, T. 2016. Spline Based Survival Model for Credit Risk Modelling, European Journal of Operational Research, 253(3), pp. 869 – 879.
[5]Bolton, C. 2009. Logistic Regression and its Application in Credit Scoring, Universidade de Pretoria, 238p.
[6]Redondo, J. M., Ortin, F. 2017. A Saas Framework for Credit Risk Analysis Services. IEEE América Latina Transactions, 15(3) pp. 474-481.
[7]Acevedo, Y. V. N., Quintero, J. F. L., Marín, C. E. M., González, C. 2016. Business Rules Model for the Automation in the Receipt of Credit Applications by Financial Institutions Based on ArchiMate. IEEE América Latina Transactions, 14(6), pp. 2801-2806.
[8]Diakoulaki, D., Mavrotas, G., Papayannakis, L. 1992. A multicriteria approach for evaluating the performance of industrial firms, Omega, vol. 20, no. 4, pp. 467 – 474, 1992.
[9]Messier, W. F., Hansen, J. V. 1985. Inducing Rules for expert system development an example using default and bankruptcy data, Management Science, 9, pp. 253 – 266.
[10]Bryant, K. 2001. ALEES: an agricultural loan evaluation expert system, Expert Systems with Applications, 21, pp. 75 – 85.
[11]Zhang, D., Zhou, X., Leung, S. C., Zheng, J. 2010. Vertical bagging decision trees model for credit scoring, Expert Systems with Applications, 37(12), pp. 7838 – 7843.
[12]Desai, V., Crook, J., Overstreet Jr, G. 1996. A comparison of neural networks and linear scoring models in credit union environment, European Journal of Operational Research, 95, pp. 24–37.
[13]West, D. 2000. Neural Network Credit Scoring Models, Computers & Operations Research,27 (11), pp. 1131–1152.
[14]Pavlenko, T., Chernyak, O. 2010. Credit risk modeling using Bayesian networks, International Journal of Intelligent Systems, 25(4) pp. 326 – 344.
[15]Gestel, T. V., Baesens, B., Garcia, J. Dijcke, P. V. 2003. A support vector machine approach to credit scoring, Journal of Banking & Finance, vol. 2, pp. 73 – 82.
[16]Belloti T., Crook, J. 2009. Support Vector Machines for credit scoring and discovery of significant features, Expert Systems with Applications, 36(2), pp. 3302 – 3308.
[17]Zhang, Z., Gao, G., Shi, Y. 2014. Credit risk evaluation using multi-criteria optimization classifier with kernel, fuzzyfication and penalty factors, European Journal of Operational Research, 237(1), pp. 335 – 348.
[18]Shanmugam, R., Johnson, C. 2007. At a crossroad of data envelopment and principal component analysis, Omega, vol. 35(4), pp. 351 – 364.
[19]Iazzolino, G., Bruni, M. E., Beraldi, P. 2013. Using DEA and financial ratings for credit evaluation: an empirical analysis, Applied Economics Letters, 20, (14), pp. 1310 – 1317.
[20]Loterman, G., Brown, I., Martens, D., Mues, C., Baesens, B. 2012. Benchmarking regression algorithms for loss given default modeling, International Journal of Forecasting, 28, pp. 161 – 170.
[21]Zhao, Z., Xu, S., Kang, B. H., Kabir, M. M. J., Liu, Y., Wasinger, R. 2015. Investigation and improvement of multi-layer per-ceptron neural networks for credit scoring. Expert Systems with Applications. 42 (7), pp. 3508-3516.
[22]García, V., Marqués, A. I., Sánchez, J. S. 2014. An insight into the experimental design for credit risk and corporate bankruptcy prediction systems. Journal of Intelligent Information Systems. 44 (1), pp. 159-189.
[23]Kao, L. -J., Chiu, C. -C., Chiu, F. -Y. 2012. A Bayesian latent variable model with classification and regression tree approach for behavior and credit scoring, Knowledge Based Systems,36, pp. 245 – 252.
[24]Abdou, H. A., Tsafack, M. D. D., Ntim, C. G., Baker, R. D. 2016. Predicting creditworthiness in retail banking with limited scoring data, Knowledge Based Systems, 103, pp. 89 - 103.
[25]Bravo, C., Thomas, L. C., Weber, R. 2015. Improving credit scoring by differentiating defaulter behavior, Journal of Operations Research Society, 66, pp.771 – 781.
[26]Steiner, M. T. A., Nievola, J. C., Soma, N. Y., Shimizu, T., & Steiner Neto, P. J. 2007. Extração de regras de classificação a partir de redes neurais para auxílio à tomada de decisão na concessão de crédito bancário. Pesquisa Operacional, 27(3), pp. 407-426.

FAQs

Machine Learning Techniques in Bank Credit Analysis? ›

The ability of machine learning techniques to use historical and existing factors to predict future trends proves useful in developing early warning systems by detecting signs of financial distress and possible changes in financial situation of the borrower.

How is machine learning used for credit scoring? ›

In contrast, machine learning credit scoring systems use traditional data (like aggregated credit scores) and alternative data (e.g., rental payments, mobile data, etc.) to identify borrower behavior patterns. Machine learning uses these learned patterns to predict the likelihood of different credit risks.

View Details ›

How is machine learning used in banking? ›

Machine learning systems can detect fraud by using various algorithms to sift through massive volumes of data. Banks can monitor transactions, keep an eye on client behavior, and log information to extra compliance and regulatory systems to help minimize overall risk when it comes to regulatory compliance.

Discover More Details ›

Which technique is used in credit risk analysis? ›

Credit risk analysis models can be based on either financial statement analysis, default probability, or machine learning. High levels of credit risk can impact the lender negatively by increasing collection costs and disrupting the consistency of cash flows.

Which type of machine learning algorithm is commonly used for credit risk classification? ›

The most effective machine learning algorithm for credit risk prediction in the provided paper is XGBoost. The paper states that the least squares support vector machine model is more effective than other classification models for credit risk prediction.

View Details ›

What algorithm is used for credit score? ›

fico score: The FICO score is the most widely used credit scoring algorithm in the US, and it is used by most lenders to determine creditworthiness. FICO scores range from 300 to 850, with higher scores indicating better creditworthiness.

How do you use AI for credit scoring? ›

Once the machine learning model is trained, it can be used for predictive analytics. When a new credit application is received, the AI system evaluates the applicant's data against the patterns learned during training. It then generates a score that predicts the applicant's creditworthiness.

Read On ›

What is the algorithm used in banking? ›

By using the Banker's algorithm, the bank ensures that when customers request money the bank never leaves a safe state. If the customer's request does not cause the bank to leave a safe state, the cash will be allocated, otherwise the customer must wait until some other customer deposits enough.

Find Out More ›

What type of machine learning is used in finance? ›

How Can Machine Learning Be Used in Finance? Some of the most widely adopted applications of machine learning in finance include fraud detection, risk management, process automation, data analytics, customer support, and algorithmic trading.

What are the 5 C's of credit risk analysis? ›

Called the five Cs of credit, they include capacity, capital, conditions, character, and collateral. There is no regulatory standard that requires the use of the five Cs of credit, but the majority of lenders review most of this information prior to allowing a borrower to take on debt.

How do banks analyze credit risk? ›

Lenders look at a variety of factors in attempting to quantify credit risk. Three common measures are probability of default, loss given default, and exposure at default.

Get More Info ›

What are the techniques of credit analysis explain? ›

This involves assessing the borrower's competitive position, market trends, and other relevant factors that may impact their ability to repay the loan. To analyze the borrower's industry and market conditions, lenders may use various tools and techniques, such as SWOT analysis, Porter's Five Forces, and PEST analysis.

View Details ›

Which ML algorithm can be used in credit scoring? ›

Various machine learning algorithms can be used for credit scoring and decisioning, including logistic regression, decision trees, random forests, support vector machines, and neural networks.

View Details ›

What are the four 4 types of machine learning algorithms? ›

As new data is fed to these algorithms, they learn and optimise their operations to improve performance, developing 'intelligence' over time. There are four types of machine learning algorithms: supervised, semi-supervised, unsupervised and reinforcement.

See Details ›

What are the machine learning techniques for credit risk evaluation? ›

The traditional machine learning models for credit risk contain Support Vector Machines (SVMs) [5], k-Nearest Neighbor (k-NN) [3], Random Forests (RFs) [4], Decision Trees (DTs) [27,28,29], AdaBoost [30], Extreme Gradient Boost (XGBoost) [31], Stochastic Gradient Boosting (SGB) [32], Bagging [33], Extreme Learning ...

See Details ›

What is the machine learning model for scoring? ›

In machine learning, scoring is the process of applying an algorithmic model built from a historical dataset to a new dataset in order to uncover practical insights that will help solve a business problem. Model development is generally a two-stage process.

Get More Info Here ›

How is machine learning used in finance? ›

ML technology is often used in finance to support investment decisions by identifying risks based on historical data and probability statistics. It can also be used to weigh possible outcomes and develop risk management strategies.

View Details ›

How is score calculated in machine learning? ›

The F1 score, a crucial metric in classification, is calculated as F1 = 2 * (precision * recall) / (precision + recall). Precision (accuracy of positive predictions) is computed as TP / (TP + FP), and recall (model's ability to capture all actual positives) is calculated as TP / (TP + FN).

Discover More Details ›

How do credit card companies use machine learning? ›

ML in the Credit Card Industry

By going through past transactions and finding patterns, ML models can suggest personalised offers and promotions for each individual customer. This means that customers get more relevant offers. ML also improves fraud detection which protects consumers against scammers and thieves.

Explore More ›