A Study On Correlation Between Web Search Data And Commercial Residential Market

In China,the real estate industry has developed rapidly since the implementation of land auction system,which has made important contributions to the economic growth of China.With house price going up,it is of great significance for consumers,developers and governments to analyze and predict the developing trend of the real estate market.In traditional research of social and economic forecasting,the data used is mostly national or regional macroeconomic one,which to a certain extent has finiteness and time lag.However,using the big data of network search engine users,whose objective is to study the correlation between consumers' search behavior and residential price and volume,to explore and to predict the trend of real estate market,is an effective research method in the information era.The research object of this paper is the new commercial residence in Guangzhou and Shenzhen.The independent variable is the processed data of search keyword in Baidu index,and the dependent variable is year-on-year index of the residential sales price and sales area abstracted from the National Bureau of Statistics and the economic database.The time span is totally 96 months from January 2011 to December 2018.This paper first summarizes the overview of China's real estate development and regulatory policies,and describes the principle and significance of network search and keyword,thus sorts out the influence factors of the real estate market and constructs a theoretical framework taking the market supply and demand relationship theory as the core.The paper selects 7 initial keywords applying the literature research method and the empirical method,and gets 139 extended keywords applying the text mining technology such as twice searching and long tail mining and so on.The search volume trend graph of Baidu Index is recognized and its massive data is processed through Matlab,and then the filtered keywords containing the temporal series information are obtained by using time relation analysis.Next,with the help of Eviews10,the ADF stationarity test,Granger causality test and Cointegration test of keyword data are carried out,then the principal component analysis is carried out by SPSS22.0,and the correlation between variables is fitted by constructing multivariate linear regression model,and the practicability of the model is proved by verification.The paper uses random forest model to predict the dependent variables.High goodness of fit shows that there is a strong correlation between the keyword index of network search and the index of housing market,and it's feasible to predict residential price and volume according to the network search data.And then,this paper makes a correlation analysis about the commercial housing market and urban characteristics of Guangzhou and Shenzhen,so as to explain the correlation between keyword data and residential market from the perspective of city.The paper finds that the correlation between sales area and network search index is stronger than that of residential price and network search index,and there are differences in network search keywords related to residential market and its correlation degree between different cities.In addition,it constructs an explanatory logical framework of the correlation between network search data and residential market from the perspective of urban development,and makes an empirical study,which provides a new perspective and some reference for the study of the regional difference of real estate in China.