House Price Prediction Using the Random Forest Algorithm on the Rapidminer Application

Fitria Fitria (Politeknik Negeri Banjarmasin)
Muhammad Syahid Pebriadi (Politeknik Negeri Banjarmasin)

Formosa Journal of Science and Technology (FJST)
Vol. 4 No. 2 (2025): February 2025

Home price prediction is an important aspect of the property industry, especially for real estate agents and potential buyers. With the advancement of technology, machine learning is used to improve the accuracy of home price estimates. One of the algorithms that is often used is Random Forest, which is able to capture complex patterns in data. However, previous research has shown that although this model has high accuracy (high R²), prediction errors are still significant (RMSE and MAE are high), indicating features that have not been fully modeled well. This research uses property datasets, preprocessing (normalization and feature selection), and applies Random Forest in RapidMiner. The model was evaluated using RMSE, MAE, and R², which showed R² = 0.78, but with RMSE = 4,757,343 and MAE = 3,200,000, indicating a large prediction difference. The most influential features are the area of the building (40%), the quality of the house (25%), and the number of bathrooms (10%).