Research On Vehicle Sales Forecasting Based On Neural Network And Time Series

China's economy has shifted from a high-speed growth phase to a high-quality development phase,and is in the process of transforming the development mode,optimizing the economic structure,and transforming the growth momentum.As an important pillar industry in China,the automobile manufacturing industry has begun to show problems such as imbalance between supply and demand and a decline in the proportion of production and sales.Therefore,automakers urgently need to accurately predict vehicle sales and apply the forecast results to production and sales decisions.The neural network is an effective nonlinear prediction method,and the time series model has good applicability to the cyclical changes of automobile sales.In the monthly sales forecast for automotive companies,the ARIMA model and the RNN neural network were used for prediction.Automobile sales have long-term characteristics,and past sales affect current sales,so LSTM-RNN is selected as the modeling algorithm in the RNN neural network.The results show that both algorithms have good fit in the prediction.In order to improve the prediction model's effect,the two models are combined according to the principle of non-negative weight optimal combination.The results show that the error is further reduced by the fusion of the two models,and the effect is better than the single model.The sales model for specific models of car can increase the production and sales rate of products and reduce the probability of overcapacity.In the model building process for a specific model of the car,using the previous experimental establishment process as a reference,combined with the characteristics of the model,the model is trained to obtain a monthly sales forecasting model that can be finally applied to a specific model of car.