|
Project properties |
|
| Title | Food substitution |
| Group | Human Nutrition and Health |
| Project type | thesis |
| Credits | 36 |
| Supervisor(s) | Dominique van Wonderen |
| Examiner(s) | prof.dr.ir. EJM (Edith)Feskens |
| Contact info | Dominique.vanwonderen@wur.nl |
| Begin date | 2023/09/01 |
| End date | |
| Description | Iron is one of the nutrients of greatest concern in vegetarian diets. Non-heme iron found in plant-based foods has a relatively low absorption (1-15%) in humans compared to heme iron found in animal-sourced foods (15-40%). To optimize the nutritional adequacy of diets, diet modelling is a quantitative technique that is commonly applied. The success of diet modelling, however, heavily depends on how acceptable proposed menu plans are to consumers. For this reason, we aim to
limit the adjustments to current consumption patterns by replacing some foods by healthier alternatives: food substitutes. There are various methods described in literature to substitute foods such as food recommender systems, word embeddings and recipe completion algorithms. Currently, we are investigating the method described by De Clercq et al. (De Clercq, M., Stock, M., De Baets, B., & Waegeman, W. (2016). Data-driven recipe completion using machine learning methods.). The aim of your thesis is to improve this method for our use case. Possible improvements may include: - Apply kernel ridge regression with a non-linear instead of linear kernel - Include food category and/or nutrient data - Develop method to estimate the quantity of substituted foods. Aim of project: Improve absorbable iron content in vegetarian diets by data- driven food substitution. Keyword(s): Diet modelling, Food substitution, Recipe completion, Machine Learning, R Programming, Data science. |
| Used skills | |
| Requirements | You have experience with R and packages dplyr and ggplot2 |