Real Estate Price Prediction Based On Web Search Data,with The Case Of Shanghai

The real estate industry is an important pillar industry of China's national economy,of which housing prices are the most critical and core elements.Traditional forecasting methods are difficult to achieve accurate forecasting of real estate prices.How to tap market information and construct more effective indicators and models to improve forecasting accuracy have become an urgent problem.With the development of big data,the gradual maturity of machine learning and natural language processing technologies,it is possible to analyze real estate prices based on web search data,which can make up for the subjectivity of index selection in the construction of traditional forecasting models and become more comprehensive.Rich influence factors can better reflect the price formation mechanism of house prices and improve the accuracy of forecasts.On the basis of sorting out the existing research results,this paper conducts in-depth research on the formation mechanism of real estate prices and analyzes the micro-behavior of the real estate market participants-real estate developers and home buyers-in the face of market competition and macro-control.As the demand side and the supply side,both of them have the process of forming psychological expectations.They will conduct network searches before economic decision-making to obtain relevant information and make investment or consumption judgments.As the Internet has penetrated into people's daily life in all aspects,people are increasingly developing information channels from traditional channels to information channels.The changing trend of Internet search data directly reflects changes in market demand and supply,and ultimately reflects At the market price of the commodity.Macroeconomic factors affect the relationship between supply and demand in real estate,and the relationship between supply and demand in real estate determines prices.Because of the delay in the changes in the behavior of participants involved in prices,the changes reflected in Internet search data are instantaneous.Therefore,this article introduces web search into the real estate market,and establishes and analyzes the antecedent-lagged relationship between Baidu Index and real estate prices.Based on the above background,this article takes real estate price as the research object,comprehensively determines the web search keywords related to real estate prices,selects its corresponding Baidu index,and screens and tests it.Based on this,it uses 10000.A 10-fold cross-validation method is used to build a price forecasting model based on web search.Further,based on the real estate prices from January 2011 to December 2017 in Shanghai,this paper conducts empirical analysis and establishes a linear regression model,random forest,and elastic network algorithms to solve and compare.The research conclusions show that:First of all,under the background of booming information technology,more and more market participants conduct information retrieval through the Internet,and the search traces of the Internet also reflect people's economic activity intentions and dynamics.The factors affecting the relationship between supply and demand in the real estate market can be reflected in the web search keywords,and there is a correlation between the two.Secondly,the price prediction model that joins the Internet search index is basically satisfactory for real estate price fitting and forecasting,realizing real-time monitoring of house prices.Among them,the randomized forest model has the smallest mean squared error(MSE),and the best fit and prediction results are obtained Finally,comparing with the existing literature,it is found that broadening the scope of keywords and increasing academic keywords can improve the accuracy of fitting and forecasting to some extent.This article based on the Internet search house price forecasting method to make up for the limitations of the traditional statistics data release relatively lag,to better address the real estate market price forecast timeliness issues,enriching the method of housing price forecasting.The forecasting framework of this paper is also applicable to other social economic indicators.It has low cost and is maneuverable.In a word,it has good theoretical and practical significance.