Housing Price Prediction Using a Hybrid Genetic Algorithm with Extreme Gradient Boosting

Predicting property prices provides a better service for customers to evaluate and estimate price movement before their purchases. Some features including OverallQual and GrLivArea, which were selected when applying GA, become important features that can influence property prices. This research proposes a hybrid Genetic algorithm combined with the Extreme Gradient Boosting algorithm to predict real estate housing prices. The proposed scheme is evaluated by Root Mean Square Error, processing time, and the number of deleted features. The proposed scheme has been compared with the sole Extreme Gradient Boosting. The experimental results show that the proposed scheme produces the smallest root mean square error value of 0.129 compared to 0.133 of the sole Extreme Gradient Boosting. Furthermore, the predicted time of the proposed scheme is much better than the sole method.