engleski

Real Estate Market Price Prediction Framework Based on Public Data Sources with Case Study from Croatia

This study uses machine learning algorithms as a research methodology to develop a housing price prediction model of apartments in Zagreb, Croatia. In this paper we’ve analyzed Croatian largest real estate ad online service njuskalo.hr. In period from April to May we’ve collected several times all ads related to Zagreb area. Each time approximately 8 000–9 000 ads were analyzed. To build predicting model with acceptable accuracy of housing price prediction, this paper analyzes the housing data of 7416 apartments in Zagreb gathered from njuskalo.hr portal.We develop an apartment price prediction model based on machine learning algorithms such as Random Forest, GradientBoostingAdaBoost and popular XGBoost algorithms. Final outcome of this research is fully functional apartment price prediction model to assist a house seller or a real estate agentmake better informed decisions based on house price valuation. The experiments demonstrate that the XGBoost algorithm, based on accuracy, consistently outperforms the other models in the performance of housing price prediction.

Real estate price prediction, Machine learning , Real estate prices, Prices seasonality , Behavioral economics , Housing price prediction model, Machine learning algorithms

nije evidentirano