Survey of Machine Learning in Credit Risk

Breeden, Joseph L.

doi:10.2139/ssrn.3616342

Cited by 13 publications

(18 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The post-hoc interpretation of the black-box models is based on the Shapley values and ALEs. We report the ALEs also for interpretable models to exploit a common basis for predictors comparison without incurring in the "p-value arbitrage" when evaluating white-box models via p-values and ML models via other criteria (Breeden, 2020).…”

Section: Resultsmentioning

confidence: 99%

Look Who's Talking: Interpretable Machine Learning for Assessing Italian SMEs Credit Default

Crosato,

Liberati,

Repetto

2021

Preprint

View full text Add to dashboard Cite

Academic research and the financial industry have recently paid great attention to MachineLearning algorithms due to their power to solve complex learning tasks. In the field of firms' default prediction, however, the lack of interpretability has prevented the extensive adoption of the black-box type of models. To overcome this drawback and maintain the high performances of black-boxes, this paper relies on a model-agnostic approach. Accumulated Local Effects and Shapley values are used to shape the predictors' impact on the likelihood of default and rank them according to their contribution to the model outcome. Prediction is achieved by two Machine Learning algorithms (eXtreme Gradient Boosting and FeedForward Neural Network) compared with three standard discriminant models. Results show that our analysis of the Italian Small and Medium Enterprises manufacturing industry benefits from the overall highest classification power by the eXtreme Gradient Boosting algorithm without giving up a rich interpretation framework.

show abstract

Section: Resultsmentioning

confidence: 99%

Look Who's Talking: Interpretable Machine Learning for Assessing Italian SMEs Credit Default

Crosato,

Liberati,

Repetto

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Breeden [19] compared some of the machine learning models for credit risk analysis like logistic regression, Support Vector Machine (SVM), K-Nearer Neighbour (KNN), Neural Network (NN), boosting models and stated that it could not be one model that can be declared as the best model. All methods have certain pros and cons for different applications and datasets.…”

Section: Related Workmentioning

confidence: 99%

A Comparative Performance Assessment for Prediction of Loan Approval in Financial Sector

Hota

2023

Preprint

View full text Add to dashboard Cite

Loans are assets that generate income in terms of interest to banks.Lending a loan to a customer creates credit and liability for the bankas well as the customer. The profit and loss of a bank depend on thecustomer’s ability to pay back the loan or not, i.e., defaulter or not.Extensive research on models for loan prediction is crucial for bankingsectors. Ensemble learning has been incorporated into the implementa-tion of many banking applications as improved machine learning modelsare continually being developed, and numerous studies have addressedthe superiority of ensemble learning. The accuracy (ACC), Receiver Operating characteristic Curve (ROC) Area Under the Curve (AUC),Kolmogorov-Smirnov Statistic (KS), Cohen’s Kappa Score (CKS), BrierScore (BS), and of ensemble algorithms, comprising bagging, boostingand Stacking, are all evaluated in this paper in context with the pre-diction of loan. Additionally, the neural network (NN), decision tree(DT), logistic regression (LR), Nave Bayes (NB), and support vec-tor machine (SVM) are five well-known baseline classifiers that areregarded as benchmarks. The experimental results affirm that ensem-ble learning performs better than individual learning. LR outperforms the baseline classifiers. Whereas the RF performs the best among theensemble approaches, followed by XGB and LightGBM, respectively.

show abstract

“…In [98], principles of machine learning methods are not clearly displayed. In [99], abundant bibliography is shown. However, the structure of the paper is not balanced.…”

Section: Existing Survey Papersmentioning

confidence: 99%

Machine learning-driven credit risk: a systemic review

Shi

Tse

Luo

et al. 2022

Neural Comput & Applic

View full text Add to dashboard Cite

Credit risk assessment is at the core of modern economies. Traditionally, it is measured by statistical methods and manual auditing. Recent advances in financial artificial intelligence stemmed from a new wave of machine learning (ML)-driven credit risk models that gained tremendous attention from both industry and academia. In this paper, we systematically review a series of major research contributions (76 papers) over the past eight years using statistical, machine learning and deep learning techniques to address the problems of credit risk. Specifically, we propose a novel classification methodology for ML-driven credit risk algorithms and their performance ranking using public datasets. We further discuss the challenges including data imbalance, dataset inconsistency, model transparency, and inadequate utilization of deep learning models. The results of our review show that: 1) most deep learning models outperform classic machine learning and statistical algorithms in credit risk estimation, and 2) ensemble methods provide higher accuracy compared with single models. Finally, we present summary tables in terms of datasets and proposed models.

show abstract

Survey of Machine Learning in Credit Risk

Cited by 13 publications

References 0 publications

Look Who's Talking: Interpretable Machine Learning for Assessing Italian SMEs Credit Default

Look Who's Talking: Interpretable Machine Learning for Assessing Italian SMEs Credit Default

A Comparative Performance Assessment for Prediction of Loan Approval in Financial Sector

Machine learning-driven credit risk: a systemic review

Contact Info

Product

Resources

About