Abstract: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.

Forecasting the Unemployment Rate by Neural Networks Using Search Engine Query Data

A Neural Network Based Forecasting Method For the Unemployment Rate Prediction Using the Search Engine Query Data

Data Mining for Unemployment Rate Prediction Using Search Engine Query Data

An Ontology-Based Web Mining Method for Unemployment Rate Prediction.

Forecasting unemployment using Internet search data via PRISM

A Neural Netwok Based Approach to Detect Influenza Epidemics Using Search Engine Query Data

Forecasting Hotel Room Demand Using Search Engine Data

Forecasting Tourist Arrivals with Machine Learning and Internet Search Index

Predicting Unemployment Claims Using Regional and Exogenous Signals: A Sparse Modeling Approach

Forecast of the Employment Situation of College Graduates Based on the LSTM Neural Network

Empirical Research on Unemployment Early -Warning Based on Regression Analysis

Neural Demographic Prediction Using Search Query.

Analysis and Forecast of China's Unemployment Rate

Forecasting Chinese GDP Using Online Data

Forecasting tourism demand with search engine data: A hybrid CNN-BiLSTM model based on Boruta feature selection

Combining Online News Articles and Web Search to Predict the Fluctuation of Real Estate Market in Big Data Context

Using Machine Learning Methods to Predict Consumer Confidence from Search Engine Data

Forecasting hourly attraction tourist volume with search engine and social media data for decision support

Scenic Spot Tourists Flow Prediction Research Based on Web Search Items

Do Internet Search Data Help Forecast Air Passenger Demand? Evidence From China’s Airports

Detecting Elevated Air Pollution Levels by Monitoring Web Search Queries: Deep Learning-Based Time Series Forecasting