Transformer Based Modeling for Nutrient Content Prediction Using Ensemble Regression

Yenugula, Anjali (2025) Transformer Based Modeling for Nutrient Content Prediction Using Ensemble Regression. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (680kB) | Preview

Abstract

The Data-driven solutions in nutrition science require models that deliver precise predictions about nutrient composition from intricate tabular datasets. A hybrid predictive framework based on TabTransformer deep learning model and Light-GBM and Random Forest enhances nutrient prediction accuracy. Traditional machine learning methods do not effectively model complex interactions between numerical and categorical features, therefore this research aims to solve this challenge. The TabTransformer component uses attention mechanisms to generate rich feature embeddings which tree based regressors utilize for predicting essential nutritional targets including energy, protein and fat. The proposed models outperform baseline models including SVR and KNN according to the comparative analysis. The results indicate that combining deep feature extraction with ensemble methods produces superior predictive accuracy which enables computational nutrition advancement and personalized dietary applications.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Gyamfi, Eric UNSPECIFIED
Subjects:	Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning Q Science > QP Physiology > Nutrition
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	27 Aug 2026 09:30
Last Modified:	27 Aug 2026 09:30
URI:	https://norma.ncirl.ie/id/eprint/9684

Actions (login required)

View Item