Introducing Network Attention To Automobile Sales Forecast

Automobile industry is one of the important pillar industry of national economy in our country, it has an extremely important role to the national economy and our country’s comprehensive strength.At present, our country is the world’s largest automobile production and consumption market, it has been also maintaining a good momentum of development.In November 2015, China’s auto production hit a high record.During the period of January to November 2015, production was 2182.39 and consumption was 21.7866 million in China, the year-on-year growth was 1.8% and 3.34% respectively.At present, the development of the auto industry is affected by multiple factors, such as the economic development level, people’s income, national policy, in addition, it also affected by the new technologies including new energy automotive and Internet automotive. Auto industry in China is facing many opportunities and challenges, so for car related companies, if they want to keep the market, they must timely and accurately grasp the market trend, make scientific and effective decisions.With the rapid development of information technology, the Internet has been widely spread in our country, the scale of Internet users increases year by year, at the same time, the regional differences are also decrease year by year. Up to December 2014, the number of Chinese Internet users reached 648.75 million, Internet penetration rate is as high as 47.9%.Among the Internet users, there are a large amount of search engine users. While in the market share of search engines, Baidu is at the absolute leading position in both PC users and mobile phone users.The popularity of the Internet has gradually changed the traditional model of consumer behavior. The Internet has become an important channel for consumers to get product information. Particularly in the auto market, the internet has obvious advantages compared with other traditional information channel. According to the investigation,77.5% of Internet users will choose the Internet as the main channel to get car information.In this paper, I firstly introduced the forecasting theory and the consumer purchase decision theory, then summarized the domestic and foreign existing research on the prediction and prediction based on web search, and their deficiency in related research. Finally I derived the methods and models in this paper. Actually, I collected information about 21 models cars from January 2011 to November 2015.The car information include sales data, network search data, car related attribute data, policy data, and macroeconomic policy data. I used the relevant knowledge of econometrics to build five models, including basic model, control model, linear combination model,time series model and time series combination model, and use many strict methods to inspect and analyst the model’s fitting coefficient,hypothetical conditions and residual, then on those basis to improve the models. Before and after the improvement, I calculated the models’predicted mean absolute error rate, compared and analyzed all kinds of model. Finally, I summarized the different influence of each variable to car sales, optimal prediction model, and the actual application value of the research in this paper.