Pengembangan Aplikasi Prediksi Harga Tandan Buah Segar (Tbs) Kelapa Sawit Menggunakan Random Forest Regressor

Fluctuations in the price of Fresh Fruit Bunches (FFB) of oil palm, influenced by Crude Palm Oil (CPO) prices, the K Index, and global market dynamics, create uncertainty for farmers and collectors when planning sales. This study aims to develop a Kotlin-based Android application to predict the average FFB price while providing a transaction-recording feature. Historical FFB price data were preprocessed and modeled using machine learning techniques, specifically the Random Forest Regressor, with linear regression employed as a baseline model. The best-performing model was integrated into the application through a Flask-based API. Firebase was used to handle authentication and data storage, while a cloud scheduler enabled automatic prediction updates. Evaluation using MAE, RMSE, and R² indicated that the Random Forest Regressor outperformed Linear Regression, generating more stable predictions that closely reflect real-world conditions. The developed application is considered suitable as a decision support system for predicting oil palm FFB prices.