Analyzing Customer Sentiments and Trends in Turkish Mobile Banking Apps: A Text Mining Study

Abstract

This study investigates customer satisfaction with mobile banking applications in Turkey through a comprehensive text mining analysis of user-generated reviews. Drawing from a large corpus of data across ten leading Turkish banks, including Ziraat Bank, İş Bank, Garanti BBVA, Akbank, Yapı Kredi Bank, Halkbank, Vakıfbank, DenizBank, QNB Finansbank, and Turkey Şekerbank, the alignment between user ratings and sentiments is explored to uncover the nuances of customer feedback. The dataset undergoes rigorous preprocessing, sentiment analysis, trend analysis, and Latent Dirichlet Allocation (LDA) topic modeling to identify prevailing themes and factors affecting user satisfaction. The methodology involves the classification of reviews into positive, negative, and neutral sentiments and the examination of trends over time to pinpoint periods of heightened dissatisfaction. The analysis is further augmented by the application of advanced machine learning algorithms, including Random Forest, Gradient Boosting Machine, and BERT, showcasing an accuracy range between 92% and 95% in sentiment classification. The results of the topic modeling are visualized through word clouds, providing a clear depiction of the dominant themes in user feedback. Trend analysis over time identifies critical periods where negative reviews surpass positive ones, often coinciding with app updates or changes in service features. The findings highlight the necessity for continuous improvement and testing of mobile banking applications to meet customer expectations effectively.