<![CDATA[Cocoa production in Desa Minanga exhibits unstable fluctuations, which directly impact the uncertainty of farmers' incomes and pose challenges in financial planning. This study aims to apply the Backpropagation Artificial Neural Network (ANN) algorithm as an instrument for predicting cocoa yields. The research data covers the period from 2019 to 2023, comprising a total of 2,980 data points. The predictive model was developed using eight main criteria: land area, number of plants, seed type, fertilizer type, pest/disease attacks, mitigation efforts, rainfall levels, and previous harvest yields. Testing was conducted using three training-to-testing data ratio scenarios: 70:30, 80:20, and 90:10. These variations were used to evaluate the model's stability and performance in identifying data patterns. Furthermore, comprehensive testing was performed on various network architecture parameters, learning rates, and target errors, utilizing the binary sigmoid activation function to assess the model's stability and accuracy in recognizing complex data patterns. The research results indicate that the optimal model configuration was achieved with a 90:10 data ratio, a 7-6-1 network architecture, a learning rate of 0.4, and a target error of 0.0001. This model achieved an accuracy rate of 98.09% with a Mean Absolute Percentage Error (MAPE) of 1.91%. The findings demonstrate that the Backpropagation ANN is effective and can serve as an alternative method for predicting cocoa yields in Desa Minanga.]]>
Copyrights © 2026
