Optimizing Library Visitor Satisfaction Analysis with Machine Learning

Abstract

In today’s increasingly digital era, libraries continue to play a vital role as centers of information, knowledge, and culture. Despite the widespread availability of online information, libraries remain essential for providing diverse resources, services, and convenient facilities. The role of libraries has evolved to meet the needs and expectations of visitors, requiring ongoing innovation in services and amenities to ensure user satisfaction. This study aims to assess the level of visitor satisfaction at UNUSA Library regarding the services provided. The research utilized questionnaire data, initially collected from 802 respondents, of which 224 valid responses were analyzed. Furthermore, this study compares the predictive performance of three machine learning methods K-Nearest Neighbor, Decision Tree, and Support Vector Machine to determine which method achieves the highest accuracy in predicting visitor satisfaction. The analysis was conducted using the Orange Data Mining application as the prediction model. The results indicate that library visitors generally report a high level of satisfaction, with certain services rated more positively than others, and that machine learning models can effectively predict satisfaction levels based on visitor feedback.