Machine Learning-Based Forecasting of Copra Prices Using Random Forest Regression in The Coconut Agroindustry Sector

Abstract

This study aims to develop a machine learning-based forecasting model for predicting copra prices in the coconut agroindustry sector using the Random Forest Regression algorithm. Copra price fluctuations are influenced by multiple interconnected factors, including production volume, product quality, market demand, transportation costs, exchange rates, and environmental conditions. Conventional statistical approaches often face limitations in modeling complex and non-linear relationships among these variables. Therefore, this study applies a machine learning approach to improve forecasting accuracy and support data-driven decision-making in the agroindustry sector.

The research employed a quantitative predictive analytics approach using a dataset consisting of 360 observations collected from several coconut-producing districts in Buton Regency, Southeast Sulawesi, Indonesia. The dataset included production, economic, market, and environmental variables related to copra price formation. Data preprocessing involved missing value checking, normalization, categorical encoding, and outlier detection. The dataset was divided into 80% training data and 20% testing data. Model performance was evaluated using R² Score, Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE). The results showed that the Random Forest Regression model achieved an R² score of 0.697, MAE of 529.50, RMSE of 707.37, and MAPE of 5.76%, indicating relatively good forecasting performance. Feature importance analysis revealed that copra quality, market demand, and moisture content were the most influential variables affecting copra price fluctuations. The findings demonstrate that Random Forest Regression is capable of modeling complex and non-linear relationships among agro-industrial variables and can serve as an effective forecasting tool for copra price prediction. This study contributes to the development of data-driven smart agroindustry systems and provides practical implications for farmers, traders, agro-industrial enterprises, and policymakers in improving commodity price stability and supply chain decision-making.