Creating a meal recommender system using the K-nearest neighbor algorithm
Date
6-2024
Pagination
75 pages
Academic Strand
Science, Technology, Engineering, and Mathematics (STEM) Strand
Adviser
John Cedric C. Gaza
Principal
Buela, Mabel S.
Abstract
The study aims to create a food recommender system that makes recommendations for a balanced meal to address the Philippines' continuous yearly rise in the proportion of overweight people. In order to do this, the researchers used open-source databases to compile detailed data on a range of food items, including their coresponding food groups, nutritional information, and serving sizes. The developed program can produce meal combinations with three food items with a balanced ratio or individually serve as a significant source of macronutrients per meal using the KNN algonthm. Through experimentation, the study generated ten outputs for each predetermined macronutrient ratio as part of the review process, resulting in 400 recommendations. Tests for four distinct preset macronutrient ratios produced results of three food items, and each food category's reduced database contained 150 randomly selected items per food group for a total of 450 food items to choose from per subset of the database. The meal presets are as follows: ideal meal ratio, low-fat, low-carb, and high-protein meal ratios. The mean total percentage error is 0.8256% for all test cases combined. For individual test cases, the percentage error for protein is 0.6029%; for carbohydrates, it is 0.5389%: and for fat, it is 1.3349%. The mean running time for all test cases is 6.75 seconds. The program's ability to generate appropriate outcomes with minimal error and fast running speed by utilizing the user's supplied macronutrient and calorie ratios is essential as it advances the development of health-promoting strategies for dietary management. Through this study, the utilization of the KNN algorithm was successfully implemented.
Language
English
LC Subject
Capstone
Location
UP Rural High School
Recommended Citation
Cabangunay, Dan Anthony G. and Matanguihan, Alec Bradley P., "Creating a meal recommender system using the K-nearest neighbor algorithm" (2024). Capstones. 125.
https://www.ukdr.uplb.edu.ph/etd-capstone/125
Document Type
Capstone