Hyper Parameter Optimization in Machine Learning For Enhancing Loan Sanction Processes

• Published in: International Conference on Computing, Communication, and Networking Technologies (Online) pp. 1 - 7
• Main Authors: Saraswathi, K., Renukadevi, N.T., Akshaya, K.G., Kanishka, S.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 24.06.2024
• Subjects: Accuracy; Bayes methods; Costs; KNN; Linear regression; Logistic regression; machine learning; Machine learning algorithms; Naive Bayes; Predictive models; Random Forest; Random forests; Reliability; Support vector machines; SVM
• ISSN: 2473-7674
• DOI: 10.1109/ICCCNT61001.2024.10724586

The main activity of banks is loan lending. A bank's earnings are mostly made up of profits. These days, loans are the most important necessity in society. Only a small percentage of the total profits go to banks. It is beneficial for people to buy luxury goods of any kind, such a house or car, and for students to budget for living and learning costs. Individuals are submitting loan applications, and banks are profiting more from lending. Assessing each applicant's information by hand is more time-consuming and subject to human mistake. To determine if a borrower will be approved for a loan, machine learning techniques and algorithms including Random Forest, Naive Bayes, KNN (KNearest Neighbor), Logistic Regression, SVM (Support Vector Machine), and Logistic Regression are employed. Grid Search is a machine learning technique for hyperparameter optimization that finds the best set of hyperparameters for a model. Grid Search automates hyperparameter modifications, which boosts productivity and lowers personnel costs. The assessment's goal is to identify the set of hyperparameters that, while accounting for accuracy, yields the highest model performance. In order to find qualified applicants, we consequently create an automated machine learning loan prediction technique. With an accuracy of 98.48%, we discovered that the Random Forest method yielded the best results based on the validation set. This model outperformed other techniques, making it the most reliable for predicting loan eligibility. The implementation of this automated loan prediction system offers substantial benefits for banks and applicants by streamlining the loan approval process, enhancing accuracy, and reducing human error. Consequently, this system significantly improves efficiency and customer satisfaction in the banking sector.