ML Based Ultra Processed Food Categorization with Limited Data

Ultra processed foods have permeated modern diets, raising concerns about their health implications. This study introduces a pioneering approach utilizing ML algorithms notably Random Forest and XGBoost, to analyze packaged food products and categorize them based on their healthiness levels. Data sourced from users included extensive nutritional information of various packaged food items. Leveraging the collected data from our responses Random Forest, XGBoost and Naive Bayes Classifiers were implemented to classify the products into the categories defined by Nutri-Score FOP Labelling system which classifies the food products into 5 categories from A (being most preferable) TO E (being worst preferable) based on the score they achieve depending on their nutritional content. We are able to obtain valuable insights for consumers and regulatory bodies facilitating informed dietary choices and addressing the challenges faced with awareness of ultra-processed foods and their consumption.