Research On Real Estate Price Prediction Based On WEB Search Data And Random Forest Model

With the rapid development of search engine technology and the popularity of the Internet,the channels for people to obtain information have changed a lot,and more and more people search for information through the Internet.Because information search is the information search behavior triggered by a certain goal,motivation and demand.Therefore,the massive web search data recorded on the Internet has become a “behavior intention database” of people.These data contain trends and laws of people's behavior,reflect the changes in social economy,and provide the necessary microdata base for the study of macroeconomic issues.At present,the real estate industry has become a pillar industry of China's national economy.It is directly related to people's employment,income,consumption and other aspects,and its healthy development is also related to the development of the national economy.In today's life,house prices are a topic that people often talk about.It is closely related to people's quality of life and happiness index.More people want to see the trend of house prices in advance so that they can make timely and accurate purchase decisions.The fluctuations affect the hearts of countless ordinary people.Therefore,the research on house prices has a very strong practical significance.Timely and accurate pre-judgment of house price trend is of great significance to consumers,real estate developers and national policy-making departments.At present,China's real estate price index also has problems such as long publication time and low timeliness.Based on this background,this paper uses the web search data that can reflect people's behavioral intentions in a certain period of time to predict the real estate price index.First of all,this article reviews the application of Web search data,forecast of real estate prices,application of random forest model,and other relevant literature,and combs the development history of China's real estate market.Then based on the theory of consumer behavior and the theory of time-delay,a theoretical framework for the correlation between web search data and real estate prices was constructed.The principles of random forests and SVMs used in this paper were described.Secondly,this article takes the Shanghai housing price index as the research object,uses text mining technology to carry out text segmentation and keyword extraction on the information on the price of the network,and determines the initial keyword library for the network search;then,the use of demand map expansion,long tail keyword expansion and other methods to determine the expansion of the keyword library;then innovatively use the time difference correlation method and RF-RFE algorithm to filter the keywords and obtain the final keyword library for the establishment of the final model.Finally,this paper combines repeated cross-validation method to optimize two important parameters of the random forest model.Using the network search data,a real estate price index prediction model based on random forest is established,and the prediction effects of the random forest model and the support vector machine model are performed.Comparative analysis.The empirical results show that the use of Web search data to predict the housing price index at least one month ahead of the official data release,has a better timeliness,the final determination of the key words can better reflect the change in the price index,and random Forest model fitting and forecasting results are better.