Recommendation System for Solaria Food Menu Based on Customer Ratings using Singular Value Decomposition
Main Article Content
Abstract
Solaria is a restaurant chain with a diverse menu, which can make it difficult for customers to choose food based on their preferences. This study aims to evaluate the factors influencing the development of a recommendation system, focusing on improving prediction accuracy. Data were collected through questionnaires from 153 respondents who rated 20 Solaria menu items. The analysis involved data preparation, matrix transformation, and duplicate data removal. The Singular Value Decomposition (SVD) method was used to reduce the large matrix into three smaller matrices, revealing latent patterns in the interaction data between customers and menu items. The model was optimized to predict ratings for items customers have not yet tried, using Mean Absolute Percentage Error (MAPE) to evaluate performance. The results show that the SVD method is effective in recommending menu items that align with customer preferences. Analysis of the influencing factors revealed a significant contribution from each factor in improving prediction accuracy. The system successfully recommended items based on preference rankings, such as Seafood Fried Rice with the highest rating and Chicken Soup Noodles with the lowest rating. This study emphasizes the importance of selecting the right factors for optimal recommendation system performance. The implementation of this system not only enhances the customer’s experience in choosing food but also provides valuable insights for Solaria to better understand customer trends and preferences.
Article Details

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.