Predicting Real-Estate valuation using Random Forest Regression.

Abstract

Prediction models in real estate have a significant role to play in telling the future of the real estate industry. They have a role to play in forecasting, which is important for investors who use the knowledge to make successful decisions. We propose a methodology using a combination of Machine learning (Random Forests), Graphic Information and different regression models. Examining real estate valuation helps to understand where people tend to live in a city. Predicting real estate valuation can help the urban design and urban politics, as it could help identify what factors have the most impact on property prices. Spot checking algorithms helped us identify a candidate to model our issue and test rapidly different regression models using spot checking. By applying this methods, have found a model that help us predict the house price of unit area in Xindian district, New Taipei, Taiwan.