Abstract:
The traditional supermarket shopping experience in Sri Lanka is plagued with inefficiencies,
particularly comparison shopping and health-conscious purchasing. Consumers spend a significant amount of time traveling from store to store to compare prices themselves, which is not just time consuming but also impractical. Additionally, the lack of personalized health recommendation tools aggravates the challenge of making informed food purchase, particularly in a country where lifestyle diseases such as diabetes and hypertension are prevalent. To meet these challenges, they developed an on-the-go grocery shopping mobile application leveraging the latest technologies such as machine learning and real-time data integration. The app employs a hybrid recommendation algorithm that combines collaborative filtering and content-based filtering to provide dynamic product suggestions based on users' preferences, purchasing history, and diet. Real-time price comparison is made possible by integrating with different supermarket APIs to offer users up-to-date pricing. The backend is built with Flask (Python) for API handling and Firebase Firestone for real-time data synchronization, while the frontend uses Kotlin and Jetpack Compose for a responsive and intuitive interface. Machine learning models were trained using
TensorFlow and Scikit-Learn, and preprocessing techniques such as normalization and feature engineering were undertaken to make accurate recommendations. Usability testing revealed that 85% of users found the interface easy to use and navigate, and only a few comments were offered to make layouts more consistent. The application was demonstrated to be scalable, supporting up to 100 concurrent users with no notable loss in performance, and could handle edge cases such as API failures and incomplete data inputs. The results indicate that the application is poised to make a tangible contribution to grocery shopping in Sri Lanka by optimizing the inefficiencies of
traditional shopping methods.
