Short-Term Forecasting Arabica Coffee Cherry Yields by Seq2Seq over LSTM for Smallholder Farmers

Coffee production is a vital source of income for smallholder farmers in Mexico’s Chiapas, Oaxaca, Puebla, and Veracruz regions. However, climate change, fluctuating yields, and the lack of decision-support tools pose challenges to the implementation of sustainable agricultural practices. The SABERES project aims to address these challenges through a Seq2Seq-LSTM model for predicting coffee yields in the short term, using datasets from Mexican national institutions, including the Agricultural Census (SIAP) and environmental data from the National Water Commission (CONAGUA). The model has demonstrated high accuracy in replicating historical yields for Chiapas and can forecast yields for the next two years. As a first step, we assessed coffee yield prediction for Bali, Indonesia, by comparing the LSTM, ARIMA, and Seq2Seq-LSTM models using historical data. The results show that the Seq2Seq-LSTM model provided the most accurate predictions, outperforming LSTM and ARIMA. Optimal performance was achieved using the maximum data sequence. Building on these findings, we aimed to apply the best configuration to forecast coffee yields in Chiapas, Mexico. The Seq2Seq-LSTM model achieved an average difference of only 0.000247, indicating near-perfect accuracy. It, therefore, demonstrated high accuracy in replicating historical yields for Chiapas, providing confidence for the next two years’ predictions. These results highlight the potential of Seq2Seq-LSTM to improve yield forecasts, support decision making, and enhance resilience in coffee production under climate change.