MealCompass: A Food Recommendation System with Machine Learning

Post-pandemic, Malaysians face “choice overload” when eating out. Additionally, the rising incidence of diabetes and obesity in Malaysia emphasizes the need for healthier eating options. To address these problems, MealCompass recommends food to users based on different user-defined criteria. Moreover, it aims to enable users to find healthier options and allow restaurant owners to provide nutritional information on food items served, which studies have proved an increase in selection of healthier choices by 13.5%. A hybrid recommendation system is proven to be more effective compared to using traditional methods alone. The hybrid recommendation system is trained on Google Colaboratory, and recommendations are shown in the application through a Flask server and Retrofit client. Waterfall model is used throughout the whole project. User feedback such as cuisine preferences, diet preferences and allergy issues, as well as the ratings of recommendations from users of the application will continuously refine and enhance the recommendations, ensuring more personalized suggestions over time. User acceptance testing among 16 respondents showed satisfaction and capability to deliver accurate and diverse recommendations. Despite these successes, limitations are noted, laying the groundwork for future enhancements, such as deploying the recommendation system to the cloud.