Feasibility Classification of Free Nutritious Meal Kitchen Partners Using C4.5 for Food Safety

Abstract

The Free Nutritious Meal Program requires a rigorous selection process for kitchen partners because kitchen quality, sanitation, clean water availability, human resources, production capacity, and food distribution are directly associated with food safety assurance. This study aims to develop a C4.5-based classification model to replicate the operational feasibility assessment rules used for evaluating kitchen partners in the Free Nutritious Meal Program based on field survey data. The dataset comprised 200 kitchen partners, including micro, small, and medium enterprises (MSMEs) and catering providers, located in Asahan Regency, North Sumatra, Indonesia, and was collected through structured field observations. Each partner was evaluated using ten assessment criteria: legal compliance, location, facilities and infrastructure, sanitation, clean water availability, human resources, production capacity, food safety, distribution, and risk history. Individual criterion scores were converted into a weighted composite score, after which feasibility labels were assigned according to an operational decision rule based on a minimum threshold score of 80 and the presence of critical failure criteria, defined as mandatory indicators that automatically disqualify a partner regardless of whether the minimum score threshold is achieved. Consequently, the class labels were not derived from independent expert audits or official institutional decisions but were generated from policy-based operational rules and used as the ground truth for model training. The dataset was divided into training and testing sets using an 80:20 ratio, resulting in 160 training instances and 40 testing instances. The results showed that 48 partners were classified as Feasible, while 152 were classified as Not Feasible. The C4.5 model achieved an accuracy of 95.00%, with 90.00% precision, 90.00% recall, and a 90.00% F1-score. The most influential predictor was the number of critical failure criteria (92.97%), followed by the clean water score (7.03%). These findings demonstrate that the C4.5 algorithm can effectively extract and replicate operational feasibility assessment rules, providing a transparent, consistent, and interpretable decision-support tool for selecting kitchen partners while supporting food safety assurance.