Comparing Linear Regression, Random Forest and Support Vector Machine for Rental House Price Forecasting in Bahrain

The real estate sector in Bahrain plays a pivotal role in its economy, with factors such as land price and house price predictions being crucial. Previous studies have addressed the importance of land price prediction in Bahrain [1]. In this study, data was collected from three famous real estate websites in Bahrain for a single day in March 2024. The dataset contains 14 attributes (such as property type, size, number of bedrooms, number of bathrooms, etc.) and 3084 instances, with 80% used for training and 20% for testing. “A comparative analysis of three data mining methods, namely Linear Regression, Random Forest, and Support Vector Machine”, for rental house price forecasting in Bahrain was conducted. Unlike previous studies, this research focuses on rental house prices in Bahrain using a diverse set of attributes. “The performance of each model was evaluated using metrics such as Mean Absolute Error (MAE) and Root Mean Square Error (RMSE)”. Linear Regression (LR) had the highest MAE value of 0.1746, while SMOreg had a lower MAE of 0.1681. The lowest MAE was 0.1385, found with Random Forest (RF). The RMSE value for SMOreg was 0.2976, which was higher than LR’s 0.2935. In contrast, the lowest RMSE value was for RF with 0.2394. This paper found the Random Forest data mining method to perform the best.