Study On Automobile Sale Forecast Method Based On Network Search Data

In recent years,with the continuous development of the social economy,people's living standard has been continuously improved,the general public has gradually increased the needs of the vehicle,and the automobile industry has become a new sunrise industry,making great contributions to the improvement of our economic strength.Moreover,automobile consumption demand has gradually become an indispensable part of life.From the psychological point of view,consumers have changed from blind consumption to rational consumption in the past,and they will choose their own vehicles according to their hobbies,economic conditions and the price of the car.With the popularity of the Internet,computers and wireless terminals have become an important way for people to get the information they need.According to the research data of China Internet Network Center,the demand for Internet is becoming more and more frequent in recent years,and about 90% of consumers search for the data they need through the network.At the same time,we can obtain consumer search data from the main search engine sites(Baidu,Google and so on).From these data,we can dig out the real psychological needs of consumers,which is of great strategic significance to the market analysis and research of an industry.The forecast of car sales is beneficial to the monitoring of automobile industry capacity,avoid overcapacity,control the growth trend of industry and promote the better development of the industry.On the microcosmic,it is beneficial for the car enterprises to formulate production marketing strategy,balance supply and demand,and optimize supply chain.Through the analysis and research of effective data in the market,this paper gets the relationship between network search index,evaluation data and automobile sales volume.First,we test the correlation between the score data,the Baidu index data and the car sales,and find that there is no correlation between the score data and the sales volume,while the Baidu index has a strong correlation with the sales data.Then according to the consumer psychology and consumption demand as the standard,from the consumer mainstream brand,select the representatives of the three high school low gear brands,BMW,Volkswagen and Geely,to do the appropriate analysis,use the web crawler tool to obtain key words data,search for key words and then synthesize the network search index.The search data in the background of the degree search engine can see the real sales of each brand in the actual,and use the links between them as a research object,and establish a data model for searching the value of heat and real car sales by the website,and the improved time series model.After that,we combine time series and regression model to get a response to the actual situation,and analyze the data.Through analysis and test results can be obtained,the Baidu search index model partly shows the relationship between the network search index and the three different brands of BMW,Volkswagen and Geely.Secondly,from the goodness of fit and prediction accuracy,the Baidu index model and the time series model have all achieved the expected effect.Third,the goodness of fit and prediction accuracy of the hybrid model are better than those of the previous Baidu index and time series models.Among them,as a representative of the low end consumption,the Geely brand's car has much difference in both the goodness of fit and the forecast precision in the two other high-end car brands,and the goodness of fit is 89.83%,and the error of the forecast is very large,and it is 8.18%.While the public is the representative of the middle end consumption,the goodness of fit is 96.14%.Around 3.31%,the prediction error is about 3%;the high end of the brand is better than the middle and low end car brand,reaching 96.81%,and the error is the smallest,only 1.88%.