<![CDATA[Background: Chili pepper production in North Sulawesi Province plays a vital role in regional food supply yet experiences frequent fluctuations due to natural and seasonal factors. Specific Background: These production instabilities have led to difficulties in market price control and agricultural planning, prompting the need for accurate predictive models. Knowledge Gap: Previous studies have compared regression and neural network algorithms in various domains, but little research has focused on local agricultural commodities such as chili peppers in North Sulawesi. Aim: This study compares the performance of Multiple Linear Regression and Backpropagation algorithms in estimating chili pepper production using statistical data from the North Sulawesi Statistics Agency (2018–2023). Results: Evaluation using R-squared (R²) and Mean Absolute Percentage Error (MAPE) shows that Backpropagation achieved R² = 0.846 and MAPE = 3.235%, outperforming Multiple Linear Regression (R² = 0.228; MAPE = 6.875%). Novelty: The study uniquely applies machine learning algorithms to a regional agricultural context characterized by nonlinear and fluctuating production data. Implications: The findings demonstrate the potential of Backpropagation as a reliable predictive tool for developing intelligent agricultural systems that support production planning and food security policy in North Sulawesi. Highlights Backpropagation shows higher accuracy than Multiple Linear Regression. Estimation uses agricultural data from North Sulawesi Province. Model supports predictive systems for food production planning. Keywords Backpropagation, Chili Production, Multiple Linear Regression, Prediction Model, North Sulawesi]]>
Copyrights © 2025
