Boosted Decision Trees for Credit Scoring

The chapter developed a tree-based method for credit scoring. It is useful because it helps lenders decide whether to grant or reject credit to their applicants. In particular, it proposes a credit scoring model based on boosted decision trees which is a technique consisting of an ensemble of several decision trees to form a single classifier. The analysis used three different publicly available datasets, and then the prediction accuracy of boosted decision trees is compared with the one of support vector machines method.

Can System Log Data Enhance the Performance of Credit Scoring?—Evidence from an Internet Bank in Korea

The credit scoring model is one of the most important decision-making tools for the sustainability of banking systems. This study is the first to examine whether it can be improved by using system log data that are stoed extensively for system operation. We used the log data recorded by the mobile application system of KakaoBank, a leading internet bank used by more than 14 million people in Korea. After generating candidate variables from KakaoBank’s log data, we created a credit scoring model by utilizing variables with high information values and logistic regression, the most common method for developing credit scoring models in financial institutions. To prove our hypothesis on the improvement of credit scoring model performance, we performed an independent sample t-test using the simulation results of repeated model development and performance measurement based on randomly sampled data. Consequently, the discrimination power of the proposed model using logistic regression (neural network) compared to the credit bureau-based model significantly improved by 1.84 (2.22) percentage points based on the Kolmogorov–Smirnov statistics. The results of this study suggest that a bank can utilize the accumulated log data inside the bank to improve decision-making systems, including credit scoring, at a low cost.

A novel multi-stage ensemble model for credit scoring based on synthetic sampling and feature transformation

Credit scoring has become increasingly important for financial institutions. With the advancement of artificial intelligence, machine learning methods, especially ensemble learning methods, have become increasingly popular for credit scoring. However, the problems of imbalanced data distribution and underutilized feature information have not been well addressed sufficiently. To make the credit scoring model more adaptable to imbalanced datasets, the original model-based synthetic sampling method is extended herein to balance the datasets by generating appropriate minority samples to alleviate class overlap. To enable the credit scoring model to extract inherent correlations from features, a new bagging-based feature transformation method is proposed, which transforms features using a tree-based algorithm and selects features using the chi-square statistic. Furthermore, a two-layer ensemble method that combines the advantages of dynamic ensemble selection and stacking is proposed to improve the classification performance of the proposed multi-stage ensemble model. Finally, four standardized datasets are used to evaluate the performance of the proposed ensemble model using six evaluation metrics. The experimental results confirm that the proposed ensemble model is effective in improving classification performance and is superior to other benchmark models.

Improved credit scoring model using XGBoost with Bayesian hyper-parameter optimization

<span>Several credit-scoring models have been developed using ensemble classifiers in order to improve the accuracy of assessment. However, among the ensemble models, little consideration has been focused on the hyper-parameters tuning of base learners, although these are crucial to constructing ensemble models. This study proposes an improved credit scoring model based on the extreme gradient boosting (XGB) classifier using Bayesian hyper-parameters optimization (XGB-BO). The model comprises two steps. Firstly, data pre-processing is utilized to handle missing values and scale the data. Secondly, Bayesian hyper-parameter optimization is applied to tune the hyper-parameters of the XGB classifier and used to train the model. The model is evaluated on four widely public datasets, i.e., the German, Australia, lending club, and Polish datasets. Several state-of-the-art classification algorithms are implemented for predictive comparison with the proposed method. The results of the proposed model showed promising results, with an improvement in accuracy of 4.10%, 3.03%, and 2.76% on the German, lending club, and Australian datasets, respectively. The proposed model outperformed commonly used techniques, e.g., decision tree, support vector machine, neural network, logistic regression, random forest, and bagging, according to the evaluation results. The experimental results confirmed that the XGB-BO model is suitable for assessing the creditworthiness of applicants.</span>

Understanding the impact of borrowers' behavioural and psychological traits on credit default: review and conceptual model

PurposeThe purpose of the study is to identify certain behavioural and psychological traits of the borrowers which have the tendency to predict the credit risk of the borrowers. And the second objective is to draw a conceptual model that reveals the impact of those traits on credit default.Design/methodology/approachThe study has adopted a systematic Literature Review approach to identify those behavioural and psychological traits of borrowers that reflect on the tendency to predict the credit default of borrowers.FindingsThe findings of this study have revealed that there are some non-financial factors, which can be looked into while granting a loan to a borrower. The identified factors can be used to develop a subjective credit scoring model that can quantify and verify the soft information (character and reliability) of debtors. Further, a behavioural credit scoring model will help in easing the assessment of those borrowers, who do not have an appropriate credit history and reliable financial statements.Practical implicationsThe proposed model would help banks and financial institutions to evaluate those borrowers who lack substantial financial information. Further, a subjective credit scoring model would help to evaluate the credit worthiness of such borrowers who do not have any credit history. The model would also reduce the biasness of subjective scoring and would reduce the financial constraints of borrowers.Originality/valueBy reviewing the literature, it has been observed that there are very few studies that have exclusively considered the behavioural and psychological factors in credit scoring. Several studies have linked the psychological constructs with debts, but very few researchers have considered it while constructing a behavioural scoring model. Thus, it can be inferred that this area of behavioural finance is still unexplored and needs attention of researchers worldwide. In addition, most of the studies are carried out in European, African and American regions but are almost non-existent in the Asian markets.

A multicriteria credit scoring model for SMEs using hybrid BWM and TOPSIS

Small- and medium-sized enterprises (SMEs) have a crucial influence on the economic development of every nation, but access to formal finance remains a barrier. Similarly, financial institutions encounter challenges in the assessment of SMEs’ creditworthiness for the provision of financing. Financial institutions employ credit scoring models to identify potential borrowers and to determine loan pricing and collateral requirements. SMEs are perceived as unorganized in terms of financial data management compared to large corporations, making the assessment of credit risk based on inadequate financial data a cause for financial institutions’ concern. The majority of existing models are data-driven and have faced criticism for failing to meet their assumptions. To address the issue of limited financial record keeping, this study developed and validated a system to predict SMEs’ credit risk by introducing a multicriteria credit scoring model. The model was constructed using a hybrid best–worst method (BWM) and the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS). Initially, the BWM determines the weight criteria, and TOPSIS is applied to score SMEs. A real-life case study was examined to demonstrate the effectiveness of the proposed model, and a sensitivity analysis varying the weight of the criteria was performed to assess robustness against unpredictable financial situations. The findings indicated that SMEs’ credit history, cash liquidity, and repayment period are the most crucial factors in lending, followed by return on capital, financial flexibility, and integrity. The proposed credit scoring model outperformed the existing commercial model in terms of its accuracy in predicting defaults. This model could assist financial institutions, providing a simple means for identifying potential SMEs to grant credit, and advance further research using alternative approaches.

Algorithmic fairness in credit scoring

The use of machine learning as an input into decision-making is on the rise, owing to its ability to uncover hidden patterns in large data and improve prediction accuracy. Questions have been raised, however, about the potential distributional impacts of these technologies, with one concern being that they may perpetuate or even amplify human biases from the past. Exploiting detailed credit file data for 800,000 UK borrowers, we simulate a switch from a traditional (logit) credit scoring model to ensemble machine-learning methods. We confirm that machine-learning models are more accurate overall. We also find that they do as well as the simpler traditional model on relevant fairness criteria, where these criteria pertain to overall accuracy and error rates for population subgroups defined along protected or sensitive lines (gender, race, health status, and deprivation). We do observe some differences in the way credit-scoring models perform for different subgroups, but these manifest under a traditional modelling approach and switching to machine learning neither exacerbates nor eliminates these issues. The paper discusses some of the mechanical and data factors that may contribute to statistical fairness issues in the context of credit scoring.

Research on Risk Assessment Method of Enterprise's Arrears Based on Block-Chain Credit Consensus

Background:At present, the power supply market has always acquiesced to the rule of ‘use electricity first, pay later’,some electricity users may delay payment or even default on electricity bills due to various reasons, causing problems such as a long recovery period for electricity bills and difficulty in debt settlement. In order to reduce this kind of phenomenon, the power supply company must understand the user's historical electricity consumption data, capital status, credit status and other information, but the establishment of such a database requires a lot of time and human resource costs.Methods:Based on the distributed storage technology of blockchain, this paper abstracts power supply companies, power users, banking financial institutions, government regulatory agencies, etc. into nodes on the alliance chain.After that,this paper introduces a credit scoring model to judge the credit rating of user information based on characteristic indicators, and select the corresponding electricity fee recovery policy after the result is obtained, so as to reduce the operating risk of the power supply company.Results:This article combines the power and energy market with blockchain technology to establish a secure and distributed credit data blockchain, and at the same time establish a credit scoring model based on expert review questionnaire data. The analysis results show that this mechanism is suitable for credit data storage and sharing in energy transactions. Conclusions:Research and analysis indicate that the credit risk assessment method of electricity transaction data proposed in this article provides a theoretical basis for the combination of electricity transaction credit risk assessment and blockchain technology,which will help improve the company's ability to assess the risk of arrears and reduce the operating risk of power supply companies.