Recognizing factors effecting the use of mobile banking apps through sentiment and thematic analysis on user reviews

We live in an age where the use of smart devices and Internet are redefining our community standards. Additionally, the pandemic Covid-19 enforced the community to use applications on smart devices for various activities. Currently, many organizations are developing their applications that are accessible through various platforms, including Windows Phone Store, Apple App Store, and Google Play. To facilitate the customer the banking sector is also providing their mobile applications for various online services. Mobile banking applications (mbanking apps) have considerably upgraded the efficiency of the banks and living standards of the people. The people can easily download applications from app stores and are permitted to leave reviews or comments on the mobile application. The sentiment analysis is an area that allows us to examine the user opinion to improve the online services. Therefore, for any organization it is of prime importance to explore and evaluate the weaknesses affecting the delivery of their online services. In this work, sentiment analysis is performed to evaluate ten (10) mbanking apps of Pakistan using valence aware dictionary for sentiment reasoning and machine learning (ML) based approaches. Performance of three classifiers through supervised ML techniques multinomial Naive Bayes, logistic regression, support vector machine, and ensemble model is compared and employed. Moreover, the thematic analysis of reviews is also performed to discover various factors as themes that affect the effectiveness of the mbanking apps by using Top2Vec Model. The results indicate that the ensemble model is best performing model with f1-score of 90%. The thematical analysis uncovers 346 positive themes like ease of use, helpful, reliable, user friendly, good aesthetics, convenience, secured and many more, whereas 441 negative themes comprise performance issue, poor updates/new version in apps, account registration issue, app crash problem, etc.