Predictive modeling of property price in Malaysia using tree-based models in machine learning

This study aims to conduct predictive modelling on property price prediction in Malaysia’s property market. The market, influenced by economic conditions, supply and demand, and government regulations, is dynamic and multifaceted; hence advanced forecasting methods are essential. Traditional regression models often fail to handle nonlinear and complex relationships. This study applies tree-based machine learning models, namely Decision Tree and Random Forest, for higher accuracy in property price prediction and to provide insight into the main determinants affecting pricing. Data was obtained from the National Property Information Centre, covering transactions across various property sectors from 2001 to 2022. Data cleaning, feature engineering, and data visualization in Tableau were performed to ensure integrity and engagement. Predictive modelling focuses on a comparison between Decision Tree and Random Forest. The objectives are to create an interactive dashboard, develop predictive models, and determine the best tree-based approach for predicting Malaysian property prices. Results indicate that Random Forest outperformed Decision Tree in accuracy, precision, recall, and F1-score. The visualisation dashboard and predictive models will assist stakeholders in policymaking, investment, and housing decisions. This contributes to a more transparent, data-driven, and equitable Malaysian property market, supporting future urban planning and economic strategies.