(IJACSA) International Journal of Advanced Computer Science and Applications, Vol. 14, No. 8, 2023 Eligible Personal Loan Applicant Selection using Federated Machine Learning Algorithm Mehrin Anannya, Most. Shahera Khatun, Md. Biplob Hosen, Sabbir Ahmed, Md. Farhad Hossain, M. Shamim Kaiser Institute of Information Technology, Jahangirnagar University, Bangladesh Abstract—Loan sanctioning develops a paramount financial dependency amongst banks and customers. Banks assess bundles of documents from individuals or business entities seeking loans depending on different loan types since only reliable candidates are chosen for the loan. This reliability materializes after assessing the previous transaction history, financial stability, and other diverse kinds of criteria to justify the reliance of the bank on an applicant. To reduce the workload of this laborious assessment, in this research, a machine learning (ML) based web application has been initiated to predict eligible candidates considering multiple criteria that banks generally use in their calculation, in short which can be briefed as loan eligibility prediction. Data from prior customers, who are authorized for loans based on a set of criteria, are used in this research. As ML techniques, Ran- dom Forest, K-Nearest Neighbour, Adaboost, Extreme Gradient Boost Classifier, and Artificial Neural Network algorithms are utilized for training and testing the dataset. A federated learning approach is employed to ensure the privacy of loan applicants. Performance analysis reveals that Random Forest classifier has provided the best output with an accuracy of 91%. Based on the mentioned prediction, the web application can decide whether the customers’ requested loan should be accepted or rejected. The application was developed using NodeJs, ReactJS, Rest API, HTML, and CSS. Furthermore, parameter tuning can improve the performance of the web application in the future along with a usable user interface ensuring global accessibility for various types of users. Keywords—Loan eligibility prediction; machine learning; ran- dom forest; K-Nearest Neighbour; Adaboost; extreme gradient boost; artificial neural network; federated learning I. I NTRODUCTION People all over the world reckon on banks to gain various kinds of financial support depending on their needs. Besides, depositing individual money it provides loans to its customers assessing different conditions and criteria. In general, banks variably provide sixteen types of loan applications [1]. In recent years, the lend-leasing industry has created significant growth increasing number of individuals seeking personal loans for various purposes. This increase in demand has led to a need for more efficient and accurate methods of loan applicant selection. Loan approval criteria defer from bank to bank. Forbes refers to the top five banks in the world providing different personal loan applications and sanctioning criteria with some common attributes [2]. Assessing those top five [3][4][5][6][7] banks, it is seen that some attributes like - credit score, social security number, loan amount, loan type, mort- gage information, employment, etc. are common. Depending on these criteria, traditional loan application processing carries forwards with manual reviews and human judgment which can be subjective and biased, leading to inefficient loan processing and higher default rates consuming a huge time in taking a decision which is a cumbersome task of the banking system. Due to human error, sometimes loans are sanctioned mistak- enly to some people who cannot repay banks’ money with interest in proper time. Moreover, banking sectors more or less face challenges with huge data management and security issues during data processing. But the use of FL in processing all the eligible loan applicants at a time is left behind. The primary motivation behind this research is to tackle the aforementioned challenges progressively, aiming to alleviate the burden on bankers in identifying loan defaulters and streamline the loan sanction process efficiently. By providing swift decisions, this research aims to support loan applicants in making informed choices that depend on the approval of their loans. Addition- ally, the research aims to expedite the loan sanctioning process, reducing the waiting time for loan applicants. The research introduces a web application developed using ML and DL algorithms for selecting eligible personal loan applicants in an FL approach to ensure security and a better data management process. Since, today’s modern world increasingly depends on ML for any type of big data analysis and prediction because of having different statistical models, and banks need more accurate predictive systems, in this research ML models are used for personal loan prediction. In a study [8], loan prediction has been done with a random forest algorithm providing better performance than a decision tree. Thereupon, in this research, the best accuracy-giving algorithm is selected among four ML and one DL algorithms for achieving better performance of data in checking the eligible personal loan applicants among all the submitted applications. The app uses data-driven approaches for analyzing vast amounts of data and making predictions about the candidates who are likely to be selected for the loan sanction. This leads to a more objective assessment of loan applicants and a reduced risk of loan defaults. And another lesson that has been found from analyzing different research on the loan prediction arena is, very few concrete systems have been developed for predicting eligible personal loan applicants ensuring the privacy of loan applicants. The key contributions of the research are: • To train and test a loan prediction dataset with four ML and one DL algorithm that has been found after the literature review. • To choose the best-performing ML algorithm among those five for loan prediction. • To ensure the privacy, security, and robustness of data processing, an FL approach will be utilized with different loan applicant selection datasets. www.ijacsa.thesai.org 1015 | Page