Feature Engineering Optimization on the Performance of XGBoost, Random Forest, and Support Vector Regression Algoritms in House Price Prediction

Abstract

As the years go by, the ever-increasing movement of house prices has become an important factor in investment decisions and financial planning to curb inflation. However, fluctuations or increases in house prices can be caused by various factors that can affect the value of house price predictions. This study aims to analyze the influence of optimization and the relationship between feature engineering and modeling in house price predictions. The research stages include data preprocessing, logarithmic transformation, feature engineering, data splitting, and optimization in determining parameters during tuning. Model performance is evaluated using the Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Mean Absolute Percentage Error (MAPE), and Determination coefficient (R-Squared) metrics. The results show that the Support Vector Regression algorithm produces the best performance with a MAE value of 274 million, an RMSE of 780 million, a MAPE of 7%, and an R-Squared of 98%. This research is expected to serve as a reference for future studies on regression model optimization, particularly in decision-making for more accurate house price predictions.