A DECISION SUPPORT SYSTEM: FOOD SECURITY AND NEURAL NETWORK MODEL

Abstract

Objective: This article aims to create a model to support decision making based on neural network for food nutrition security in São José dos Ramos, in the interior of Paraíba. Methods: This technique was used to create a model, through implicit patterns in layers of 10 variables on the quantity and variety of foods from 181 families in the municipality under study, using as classifier the Multilayer Perceptron Weka software. Results: Once the model was developed, there was an excellent standard classification in the distinction between food security and food insecurity, although confusion in the classification among food insecurity levels was verified. The choice for relevant variables in the implementation of neural networks must be performed carefully, since the inclusion of variables non relevant to the problem under study may affect the neural network performance, as well as the classification error. Conclusion: The model proposed here may be conducted as a support tool for the diagnosis of food insecurity, providing useful answers to the management of programs to combat hunger. DESCRIPTORS: Food security. Artificial Intelligence. Medical Informatics.