A review of Artificial Intelligence approach for credit risk assessment

Every day, each bank around the world has to analyze many credit applications from its customers and prospects, individuals, professionals, or companies. Banks develop their rating system based on different parameters but most of them do not take benefit of the tremendous set of Big Data available and gathered continuously. To extract valuable information, Big Data analysis (BDA) and artificial intelligence (AI) lead to interesting applications for the banking industry such as segmentation, customized service, customer relationship management, fraud detection, credit risk assessment, and in all back, middle, and front office missions. This article presents the benefit of artificial intelligence for credit risk assessment. A state of art for the actual research advance is discussed concerning this specific item. To handle this review, we first focused on the keywords to capture and analyze the available articles of experts. We limited the period from 2016 to 2021 to skim the recent advances. Researchers have explored different methods with feature selection, classification, and prediction. Algorithms of Data mining, machine learning (supervised and unsupervised), and deep learning (artificial neural networks) are very different and tackle various aspects to be explored. With these advances, banks can become smart and propose a better and quicker service while preserving themselves from losses due to credit defaulters. Support vector machine, Catboost, decision tree, and logistic regression have delivered interesting results according to the studied researches.