Machine learning techniques for credit risk evaluation: a systematic literature review

Bhatore, Siddharth; Mohan, Sriram; Reddy, Y. Raghu

doi:10.1007/s42786-020-00020-3

Cited by 72 publications

(31 citation statements)

References 141 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Previously many studies on credit risk modeling have shown the good performance of their methodologies with class‐balanced datasets. More specifically, 4,7,12,17 have proven the good SMOTE potential for tackling this task's issue. Therefore, before implementing the neural network architecture, we also balanced data‐3 in the proportion of 46:54 of defaults: nondefaults using the Gaussian‐SMOTE 42 .…”

Section: Proposed Methodologymentioning

confidence: 99%

See 1 more Smart Citation

A Gaussian process‐based approach toward credit risk modeling using stationary activations

Mahajan

Nayyar

Raina

et al. 2021

Concurrency and Computation

View full text Add to dashboard Cite

The task of predicting the risk of defaulting of a lender using tools in the domain of AI is an emerging one and in growing demand, given the revolutionary potential of AI. Various attributes like income, properties acquired, educational status, and many other socioeconomic factors can be used to train a model to predict the possibilities of nonrepayment of a loan or its chances. Most of the techniques and algorithms used in this regard previously do not submit any attention to the uncertainty in predictions for out of distribution (OOD) in a dataset, which contributes to overfitting, leading to relatively lower accuracy for predicting these data points. Specifically, for credit risk classification, this is a serious concern, given the structure of the available datasets and the trend they follow. With a focus on this issue, we propose a robust and better methodology that uses a recent and efficient family of nonlinear neural network activation functions, which mimics the properties induced by the widely‐used Matérn family of kernels in Gaussian process (GP) models. We tested the classification performance metrics on three openly available datasets after prior preprocessing. We achieved a high mean classification accuracy of 87.4% and a lower mean negative log predictive density loss of 0.405.

show abstract

Section: Proposed Methodologymentioning

confidence: 99%

“…Experimentally, they managed to find that merging two stages achieves good performance. Bhatore et al 17 reviewed a total of 136 studies on credit risk evaluation. The authors presented a well‐structured analysis of these studies and some good insights on various techniques and algorithms used for the past task.…”

Section: Background and Previous Workmentioning

confidence: 99%

A Gaussian process‐based approach toward credit risk modeling using stationary activations

Mahajan

Nayyar

Raina

et al. 2021

Concurrency and Computation

View full text Add to dashboard Cite

show abstract

“…Further research has been conducted on soil water resources, and significant progress has been made. In addition to recognizing soil water as a natural resource, it was proposed that "soil water is the most important natural resource" from the viewpoint of its role in the interchange of land water and environmental factors and agricultural production, which was recognized by scholars at home and abroad [7]. While the production of mariculture is increasing, the environmental problems caused by it are gradually coming to the fore, mainly in two aspects.…”

Section: Status Of Researchmentioning

confidence: 99%

High-Performance Computational Early Warning Analysis of Agricultural Economy Relying on Binary Fuzzy Cluster Analysis Algorithm

Tang

2021

Journal of Mathematics

View full text Add to dashboard Cite

In this paper, a binary fuzzy cluster analysis algorithm is used for an in-depth study and analysis of high-performance computational early warning in the agricultural economy. The definition of interval type-two fuzzy set and its operation are summarized. Considering the uncertain information that will be encountered in the evaluation process, this paper constructs the evaluation model of rural informatization construction performance based on interval type-two fuzzy numbers, which improves the accuracy of the evaluation results. The fuzzy clustering centres are modified according to the nature of the fuzzy clustering centre matrix, and the optimal fuzzy clustering centres and optimal fuzzy clustering division matrix with consistent order are solved. Using the level eigenvalues to find out in the abundant water period, there are 4 monitoring sections of water quality evaluation results for still clean and 5 monitoring sections of water quality evaluation results for slight pollution. In the flat-water period, there are 2 monitoring sections of the water quality evaluation results that are still clean and 7 monitoring sections of the water quality evaluation results are slightly polluted. In the dry water period, the water quality evaluation results of the 9 monitoring sections are slightly polluted. The results and the use of integrated pollutant index evaluation method results are consistent, indicating that the use of fuzzy clustering model for water quality evaluation is practical and effective. Its functional structure includes management business, innovation business, service business, production business, and operation business functions, in addition to the design of the collection and storage scheme based on Hadoop data warehouse, to achieve accurate matching of information of science and technology agricultural services in the whole agricultural industry chain. Finally, the implementation and maintenance of the cloud-based science and technology agricultural service information system are deployed, and the operational effects of its information system-related functions are demonstrated using prototype design. This paper constructs a comprehensive information system for science and technology agricultural services based on cloud technology that integrates management, innovation, service, production, and operation, which meets not only the needs of traditional science and technology agricultural service information systems, such as expert consultation and training demonstration, but also other needs spawned by the development of agricultural modernization.

show abstract

“…The subprime mortgage crisis during 2007–2010 was due to the inefficient and low accurate credit scoring methods ( Bhatore et al, 2020 ). In order to reduce potential non-performing assets and improve the efficiency of credit risk control, more reliable credit scoring approaches are urgently demanded.…”

Section: Introductionmentioning

confidence: 99%

“…Although these techniques can be applied to credit risk assessment, they can be further improved. In the last two decades, there has been a growing approaches proposed in the field of machine learning that can handle large amounts of data yet guarantee good accuracy ( Bhatore et al, 2020 ). For example, machine learning techniques such as Bayesian networks, decision trees, and support vector machines have been widely applied to user credit assessment.…”

Section: Introductionmentioning

confidence: 99%

A novel method for credit scoring based on feature transformation and ensemble model

Feng

Lin

et al. 2021

PeerJ Computer Science

View full text Add to dashboard Cite

Credit scoring is a very critical task for banks and other financial institutions, and it has become an important evaluation metric to distinguish potential defaulting users. In this paper, we propose a credit score prediction method based on feature transformation and ensemble model, which is essentially a cascade approach. The feature transformation process consisting of boosting trees (BT) and auto-encoders (AE) is employed to replace manual feature engineering and to solve the data imbalance problem. For the classification process, this paper designs a heterogeneous ensemble model by weighting the factorization machine (FM) and deep neural networks (DNN), which can efficiently extract low-order intersections and high-order intersections. Comprehensive experiments were conducted on two standard datasets and the results demonstrate that the proposed approach outperforms existing credit scoring models in accuracy.

show abstract

Machine learning techniques for credit risk evaluation: a systematic literature review

Cited by 72 publications

References 141 publications

A Gaussian process‐based approach toward credit risk modeling using stationary activations

A Gaussian process‐based approach toward credit risk modeling using stationary activations

High-Performance Computational Early Warning Analysis of Agricultural Economy Relying on Binary Fuzzy Cluster Analysis Algorithm

A novel method for credit scoring based on feature transformation and ensemble model

Contact Info

Product

Resources

About