An Explainable AI framework for credit evaluation and analysis

Abstract

Loan Facility is a profitable venture for the banking industry and can render great financial support to the beneficiary. The Global banking systems with secured private cloud are making the service reachable around the world around the clock. Loan acceptance and disbursal are governed by the protocol of the banks with the highest degree of privacy and integrity. As per the report of Experian, the loan acceptance rate of the banks has been reduced to 61%-70%, and it is further reduced to 50% post -pandemic since there is a huge financial setback and a higher rate of defaulters. The banks are not in a position to explain the reasoning behind the rejection since the rejection further diminishes the customers' credit scores. With the parallel improvements in Industry 5.0, futuristic banking could evolve around Non Fungible Tokens (NFT) integrated Explainable AI (XAI) framework which can interact with the customer through the human -machine interface in the meta -verse. For such a kind of system, the proposed work could be a driving application which provides explanations for the loan rejection, with the Random Forest integrated XAI framework that provides the reasons for acceptance and rejection of the loan. The proposed Random Forest -based approach rendered the highest accuracy, sensitivity and specificity of 0.998, 0.998, and 0.997, respectively. The LIME and SHAPELY Explainers provide explanations with local and global surrogates of various parameters on the features.