<![CDATA[EatAja is a startup company in Indonesia that provides solutions in order in ordering food at mobile app-based restaurants. The variety of menu from various restaurants will make it confusing for users to choosing. Due to this problem, restaurants would be displaying the best sellers’ menu. However, there is not solution for users who eat according to their taste. For restaurants, promoting specific menus based on users’ taste is quite challenging because of users have preferences by themselves and unavailability about that information. Since many users may have similar food preferences, recommender system is a must-have feature to be implemented in such applications that involve data from many users. In this research are using memory-based collaborative filtering method to check a similarity between users’ orders. By using real order history data from EatAja combined with generated auxiliary data to implicitly find customers’ ratings towards menu they have been ordered, the recommender system obtains Mean Absolute Error (MAE) 0.96823 with the best accuracy is 99.03%. The result of the recommendation system can be applied to the application to be able to increase sales to the restaurant as a suggested menu.]]>
Copyrights © 2021
