iOS Application User Rating Prediction Using Usability Evaluation and Machine Learning
Summary
This dissertation presents a novel approach to predicting iOS application user ratings based on usability evaluation metrics and machine learning techniques. The research bridges the gap between traditional usability evaluation methods and real-world user satisfaction as expressed in App Store ratings. By analyzing the relationship between specific usability factors and user ratings across multiple application categories, the work develops predictive models that can help developers anticipate how users will perceive and rate their applications before release.
Key Insights
- Specific usability attributes have varying levels of influence on user ratings depending on the application category and target user demographic
- A combination of expert-based usability evaluation and automated static analysis provides powerful predictive features for machine learning models
- The research identified five key usability dimensions that most strongly correlate with positive user ratings: interface aesthetics, navigation intuitiveness, error prevention, consistency, and feedback quality
- Traditional usability metrics alone account for approximately 65% of the variance in user ratings, with the remaining variance attributed to non-usability factors such as content quality and value perception
- The proposed predictive models achieved 78-84% accuracy in anticipating user rating ranges across various application categories
Impact
This dissertation has significant implications for iOS application development practices, offering developers a framework to proactively address usability issues that most impact user satisfaction. By enabling the prediction of potential user ratings during the development cycle, the research helps teams prioritize improvements that will have the greatest effect on App Store success. Several mobile development companies have adopted elements of this approach to enhance their quality assurance processes and improve their competitive position in the App Store marketplace.
Access
The full dissertation is available at ÉTS - Institutional Repository.