Classification of Book Borrower Patterns Based on Profession Using the Naïve Bayes Algorithm

Abstract

Libraries face challenges in understanding book lending patterns to optimize services and collections. This research classifies borrowing patterns based on profession using the Naive Bayes algorithm, with data from 4476 transactions at the Sidoarjo Library Service in 2023. Attributes include profession, book category, and reading interest level. The study involved data collection, preprocessing, processing with Gaussian and Multinomial Naive Bayes, and model evaluation. Testing across data ratios (90:10, 80:20, 75:25, 50:50) revealed that Gaussian Naive Bayes achieved the highest accuracy of 97% in a random dataset scenario. Findings indicate students, university students, and housewives dominate the high reading interest category, while doctors and researchers exhibit lower interest. This research highlights the value of data-based analysis in supporting library management, offering insights for optimizing collections, tailoring services to professional needs, and enhancing literacy programs. It serves as a foundation for leveraging data mining to address modern library management challenges.