Price prediction of fresh fruit bunches in Sepintas using predictive analysis

Abdul Hakim, Muhammad Azim (2026) Price prediction of fresh fruit bunches in Sepintas using predictive analysis. [Student Project] (Unpublished)
Abstract

Fresh Fruit Bunches (FFB) pricing is a key issue for collection centers and smallholders. Price changes can influence procurement planning, harvesting choices, and expected income. At MAH Sepakat Enterprise in Sepintas, the current pricing method relies heavily on daily price references obtained through the InfoSawit app. Historical transaction records are mainly stored for documentation and are not fully used for analysis. This project aimed to use predictive analytics for FFB price prediction and to display the results on an interactive dashboard. The project followed the CRISP-DM methodology, which includes Project Planning, Business Understanding, Data Understanding, Data Preparation, Modeling, Evaluation, and Deployment. Historical transaction records from June 2024 to November 2025 were collected, cleaned, formatted, and integrated, resulting in 2,179 records for analysis. Three regression algorithms, Random Forest, Decision Tree, and Linear Regression, were developed using RapidMiner and assessed using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and R Squared (R2). Random Forest showed the best performance, recording the lowest MAE of RM66.476 and the highest R2 value of 0.963. It was chosen to estimate FFB prices for 2026. The prediction results and insights from historical transactions were shared through a Power BI dashboard. This dashboard shows FFB purchase activity, price trends, delivery patterns, smallholder contributions, and estimated future prices. Experts reviewed the dashboard and found it useful and clear. However, they suggested improvements for user guidance, chart explanations, visual clarity, and interface design. Overall, the project offers MAH Sepakat Enterprise and local smallholders a better way to examine FFB price trends and make planning decisions. Future enhancements could include longer historical data coverage, more external factors, and the exploration of other machine learning or time-series techniques.

Item Details
Edit Item
Edit Item
Downloads & Files
[thumbnail of 149519.pdf]
Text
149519.pdf
Download (95kB)
Location & Physical Holdings
Digital Copy
Physical Location
Item Status
Indexing & Metrics