<![CDATA[House price prediction has become an important need in the property sector because selling prices are influenced by various factors, such as building quality, living area size, garage capacity, and physical property conditions. The complexity of relationships among these variables makes machine learning approaches, particularly Random Forest Regression, relevant for developing accurate and stable prediction models. This article discusses the optimization of Random Forest Regression for house price prediction through feature selection and hyperparameter tuning. The dataset used in this study is the House Prices: Advanced Regression Techniques dataset from Kaggle, with SalePrice as the target variable. The analysis process includes data exploration, data preprocessing, feature selection, train-test data splitting, Random Forest model development, and hyperparameter optimization using Randomized Search. Model performance was evaluated using R-Squared, Root Mean Squared Error (RMSE), and Mean Absolute Error (MAE). The results show that the optimized model achieved an R-Squared value of 0.8927 and an RMSE value of 28,687.08. The most dominant features influencing house prices were OverallQual, GrLivArea, and GarageCars. Although the performance improvement after optimization was relatively modest, Random Forest proved capable of producing stable predictions and can be used as a decision-support tool in the property business. However, the model still has limitations in predicting extreme prices in the luxury housing segment.]]>
Copyrights © 2026
