An Empirical Study Of Automobile Sale Forecast Under The Background Of Big Data

Generally, automobile industry is the mainstay of national economy,accurate prediction of theautomobile scale not only means great significance to automobile manufacturers and dealers,but also hasinfluence on the growth of automobile market and the development of domestic economy.China’s economy has growing in a dizzying pace since the Reform and Open-up,which leads tothe change of domestic consumption demand.Instead of “Demanding for Housing”,people trend todemand for automobile vehicles.Consumers would access to the automobiles’ information based on theirown preference,price and quality of vehicles,and economic conditions before buying automobiles.Web technology and Internet have completely changed the traditional modes that consumersaccess to information,which makes consumers prefer to acquire data and information viaInternet.According to statistics provided by China Internet Network Information Center (CNNIC),80%ofthe Netizens obtain information via search engines.Web search data shows Netizens’ focuses anddemands,reflects the trends of market entity,which provide the necessary data for the research of socialeconomical behaviors.This paper takes Cherry,Volkswagen and BMW respectively from low-market,mid-market and high-market as examples,establishes a theoretical model about automobile buyingbehaviors and explains the correlationship between the web search data and automobile sale,according toconsumers’ Internet searching records from Baidu.The mathematic methods that author used are:Combination Weighting, Stepwise Synthesis, Cointegration Test and Granger Causality Test.Data analytic result shows this model is more reliable and accurate than traditional ways:Cherry,from the low-market,is not that satisfying with the Goodness of Fit at70.0%and Mean AbsoluteError at8.51%; Volkswagen’s result is better than Cherry whose Goodness of Fit is95.2%, Mean AbsoluteError is4.85%;BMW,with Goodness of Fit at97.7%, Mean Absolute Error at2.78%, is the best amongthree brands.It is obvious that the forecasting accuracy of this model improves from low-market tohigh-market.This model can be used to automobile scale real-time monitoring,whose result can be onemonth ahead to traditional statistics.Although the relevance of research on the social web search data and economy is still in itsinfancy, but with the rich and comprehensive development of information technology and network searchdata, the research methods in the study of the theory will form a set of complete theoretical system, but also have broader practical applications, such as: real estate industry, tourism and other industries.