Web-Based Free-Range Chicken Farm Productivity Prediction System Using Random Forest Algorithm

Abstract

This research is motivated by the process of predicting native chicken egg production, which is still done manually based on farmer estimates, potentially producing inaccurate predictions and hampering production management efficiency. This study aims to develop a web-based native chicken farm productivity prediction system using the Random Forest algorithm. The study uses a quantitative approach with historical egg production data from one farm over six years. The dataset consists of 250 data with variables such as number of chickens, feed, temperature, humidity, age, and egg weight. Because it uses historical data, the study does not specify a population and survey sample. Data were collected through observation, interviews, and literature studies, then analyzed through data selection, preprocessing, data division 80% training and 20% testing, Random Forest Regression modeling, and evaluation using MAE, MSE, RMSE, and R². The results showed MAE 0.6702, MSE 1.2712, RMSE 1.1275, and R² 0.9621. In conclusion, the web-based system with Random Forest is able to produce egg number predictions with good performance and can be used to support evaluation and production planning for free-range chicken farms.