The real estate price is of paramount importance in both economic and social fields. It is a key indicator of the operation of real estate market and its prediction is essential in the decision-making process of both average people and official governments. Past researchers on this topic have already proposed several prediction methods including linear regression models, nonlinear regression models and machine learning models. Nevertheless, those models have generally neglected the impact of human behavior, which we believe is a significant factor of the real estate price prediction. What’s more, past studies have shown that news sentiments could improve the prediction performance of real estate price. Search engine query data were studied to reflect web users’ behavior by analyzing the frequency of words searched by online users. Researchers have already used the news sentiments and query data for prediction, respectively. But none have combined them together as an integrated model. In this paper, we propose an integrated method that throws new light on the prediction of real estate price in China by integrating these two factors into the forecasting model. In our method, we extract sentiment series from both news data and search engine query data by adding weights to original sentiment series that are produced by news data alone. Then both the weighted series and original ones are used as inputs of several well-acknowledged data mining models, including SVR, RBFNN and BPNN, to produce prediction results.

To validate the integrated model, we apply it to four representative cities in China respectively, and compare the results produced by the integrated model using weighted inputs with non-integrated ones using original inputs. The results show that for every one of the four cities, the integrated model generally leads to lower prediction errors than the non-integrated ones. This not only validates the model’s accuracy and universality, but also proves the hypothesis that human searching behavior as a strong impact in typical Chinese cities’ real estate market and can enhance the prediction accuracy of real estate prices.