Performance Analysis of Machine Learning for Food Fraud Prediction

Estrella Condor J.D.; Pérez F.A.F. · Communications in Computer and Information Science (2023) · DOI: 10.1007/978-3-031-35445-8_19

Resumen

Food fraud is the set of practices based on adulteration, intentional manipulation of food or counterfeiting, which are carried out to obtain an economic benefit. They are considered criminal actions, as they are punishable by law, are harmful to society and are morally reprehensible. One of the solutions to prevent the entry of fraudulent foods is their early detection. The study focuses on identifying two types of offenses, either the presence of improper sanitary certifications in a food or verifying whether the product belongs to an illegal or unauthorized import. To achieve this objective, the RASFF (Rapid Alert System for Food and Feed) data set will be used, which is a system where the European Union food and feed control authorities exchange information on the different risks that have been detected. The use of information from the RASFF model can be useful for predicting the type of fraud they are committing. Several techniques already implemented will be used to compare the results. Naive Bayes model, logistic regression, multilayer perceptron, decision tree, random forest and SVM. The best model was the multilayer perceptron.

Instituciones: Universidad Nacional Mayor de San Marcos