Research On Car Sales Forecasting Model Based On Baidu Index And BP Neural Network

The company's strategic decision-making should be based on market demand.One of the most important indicators that can truly reflect the market demand is the product sales.Therefore,the timely and accurate forecasting of car sales has important guiding significance for the future development planning of the company.At present,the competition in China's auto market is very fierce.The major automobile manufacturers have invested a lot of funds and technologies to occupy the market,and new models are continuously being introduced into the market.This can provide consumers with more abundant purchase options.On the other hand,it also makes it more difficult for auto companies to forecast the product sales.In the context of the rapid development of the Internet,the ways and habits of consumers accessing information have changed.People's decision-making behaviors are reflected on the Internet.Using the real-time Internet search data to forecast sales provides a new research perspective for enterprises.This thesis firstly establishes a theoretical framework for the relevance of consumer Internet information search and product sales.After that,based on the data left by the auto consumers' use of Baidu search engine to search for information in the decision-making process,a keyword database is established in which the word is of high-relevance and time-leading with the car sales.Then a car sales forecasting model is established based on BP neural network.Based on this,taking a specific auto model as an example,monthly sales of a model automobile can be forecasted in one month.Finally,the forecasting results of this thesis and the traditional ARIMA and multiple linear regression models are compared.The comparison results show that the model forecasting error proposed in this paper is smaller.In addition,the basic model is improved based on the principal component analysis,practice has proved that the improvement is effective.The car sales forecasting model based on Baidu index and BP neural network proposed in this thesis is a new method based on real-time Baidu index data and combined with traditional artificial neural network prediction.The forecasting object of this model is a specific model,the forecasting is more detailed and has better fitting effect and prediction accuracy.It can provide a new idea for the sales prediction of auto companies,and can also be applied for reference by other industries.