Discussion

The evaluation results demonstrate that our meal-planning assistant effectively streamlines meal selection and grocery shopping for most users. The integration of Firebase Firestore, FastAPI, and Google Maps API enabled a seamless experience, with meal recommendations and shopping lists tailored to user preferences. Compared to existing meal-planning appli- cations such as Mealime and Paprika, our system offers a higher degree of personalization and convenience by incorporating real-time store location data. While AI-driven recommen- dation systems, such as IBM Watson Meal Planner, provide personalized meal suggestions, they lack the direct integration with grocery stores and automated shopping list generation that our system provides. Despite these advantages, some challenges remain. Users in less densely populated areas experienced difficulty finding relevant grocery store recommendations due to limited store data, a limitation also observed in prior studies on location-based retail applications. While the system aimed to optimize cost efficiency, the absence of real-time store pricing some- times led to inaccurate cost estimates, a common issue in online grocery shopping platforms that rely on static pricing models. Additionally, some users expressed interest in more flex- ible meal customization options, such as ingredient substitutions and dietary-specific meal plans, indicating a need for greater adaptability in future iterations. To further improve the system, several enhancements can be considered. Implementing real-time pricing APIs from grocery retailers could improve cost estimation accuracy, ad- dressing a key limitation of the current approach. Expanding the ingredient substitution feature would provide users with greater flexibility in adjusting their meal plans based on availability and dietary preferences. Additionally, increasing store data coverage in rural areas through partnerships with regional grocery chains could enhance the system’s usabil- ity in underserved locations. Another promising direction is improving the transparency of recommendation logic, allowing users to understand why certain meal suggestions are made, which could improve trust and user engagement. Overall, the SmartCart system demonstrates strong potential as a practical solution for meal planning and grocery shopping. While it effectively personalizes meal recommendations and optimizes grocery store selection, addressing the identified limitations in future iter- ations will enhance its reliability and user satisfaction. Future work should also explore machine learning-driven optimization techniques for balancing price, distance, and ingre- dient availability in real time, further refining the system’s ability to meet user needs.