Mobile banking applications are inevitable for users in modern digital banking, but overall user satisfaction with the app is often affected by User Interface (UI) and User Experience (UX) issues especially emerging country like Sri Lanka. Existing research has mainly focused on sentiment analysis, neglecting the categorization of UI and UX concerns. This study aims to bridge this gap by using sentiment analysis and topic modeling techniques to determine either UI or UX improvements. This research employed a model capable of determining whether user concerns are related to UI or UX aspects. The dataset collected through web scraping from the Google Play Store, which includes user reviews from three popular banking apps in Sri Lanka such as X, Y and Z, was cleaned using preprocessing techniques. To develop a strong classifier for identifying UI and UX issues in negative user reviews, different machine learning models were created and experimented with, including K-Nearest Neighbors (KNN), Random Forest, Extreme Gradient Boosting (XGBoost), Categorical Boosting (CatBoost), and Artificial Neural Networks (ANN). The ANN model showed the highest accuracy in predicting UI and UX issues, and Locally Interpretable Model-Agnostic Interpretations (LIME) was used to validate its predictions. Through thematic feature comparison, the study provides new insights into the importance of addressing common application issues related to these features, such as app performance, app updates, error recovery, general usability, login process, and transaction functionality for each banking application, obtained by considering keywords derived from the negative user reviews.