PREVISÃO DE CUSTOS DE PRODUÇÃO DE FRANGOS: UMA APLICAÇÃO DO MODELO CNN-LSTM

Abstract

Poultry farming, one of the main components of agribusiness in Paraná, contributes significantly to the state economy. The knowledge, by the producer from Paraná, of the production costs of broiler chickens is essential to save resources, increase efficiency and guarantee the sustainability of their business. In this context, the main objective of this work was to propose, to predict the production costs of broiler chickens in the state of Paraná, a hybrid multivariate CNN-LSTM (Convolutional Neural Network - Long Short-Term Memory) model. The database, made available by EMBRAPA (Brazilian Agricultural Research Corporation), presents a series of monthly costs between January/2010 and April/2024. A detailed performance comparison of the CNN-LSTM model with other machine learning models was performed. Results obtained from the models were compared using the metrics RSME (Root Mean Squared Error), MAPE (Mean Absolute Percent Error) and MAE (Mean Absolute Error). It was verified, for a 12-month horizon, that the CNN-LSTM forecasting model provided reliable estimates for chicken production costs.