Machine Learning-Based Nutritional Prediction of Recipes Using Structured Ingredient Data

As more recipe and nutrition data becomes available, machine learning can now automatically analyze the nutritional value of food recipes.Manually judging whether a recipe is healthy or nutrient-dense can be slow and tough especially when you're reviewing lots of recipes.This study presents a machine learning approach to multi-label classification, helping identify the nutritional profile of recipes.We worked with 725 recipes, cleaning the data, filling in missing info, prepping ingredients, picking key features, and building new ones to make them useful.The final dataset had 15 key attributes, and ingredient info was turned into numbers to ready it for training.The study focuses on six key nutrition categories: Healthy, Protein_Rich, Fibre_Rich, Calcium_Rich, Iron_Rich, four machine learning algorithms-Decision Tree, Random Forest, Support Vector Machine (SVM), and XGBoost were trained and evaluated using a dataset split of 580 samples for training and 145 for testing.XGBoost came out on top with 75.68% accuracy, while SVM scored the best F1-0.5760.The results show machine learning can help automatically sort recipes by various nutritional labelsbut how well it works varies by category.