Research On Forecast Of Sales Of Automobile After-sales Service Products Of A Company

With the rapid development of the current economy and the diversification of customer needs,not only traditional sales enterprises are faced with rapid response to customer needs and shorten the time of commodity logistics,but also online e-commerce enterprises are faced with such response problem.Under the current new market situation and competitive pressure,sales forecast is very important for enterprises.Through effective accurate sales forecast,sales enterprise to the arrangement of the timely and accurate purchase quantity,as well as ready inventory goods,for a rainy day,also want to make sure not too much inventory resulting in Cun backlog,in turn,causes loss of enterprise,the most important thing to ensure timely response to customer’s requirement of improving customer satisfaction,to open the market reputation.Sales forecasting is the current enterprise research focus,although there are many models and methods applied to the sales forecast,but each enterprise,industry sales characteristic is inaccurate and study specific problems,do not have universal applicability,so the research on the perspective of A company,we are going to build A higher prediction accuracy,and for A enterprise sales feature model is used to sell commodities demand forecasting.The main research work of this paper includes:Firstly,by studying the current domestic and foreign literatures,the characteristics of various types of prediction methods and models are analyzed and compared.According to the characteristics of the goods sold scheduling,first Xu Na the ARIMA model is used to study variation law with time history data to predict future demand data,the ARIMA model is a linear model based on time series,this method need not to study the factors influence the sale of goods,only focus on the historical data of change over time,can use in many predict scenarios,but the ARIMA model is a simple linear model based on time series data,not fitting the nonlinear part of the time series,forecasting accuracy is not high.Therefore,based on the ARIMA model,this paper uses the mixed model to improve the accuracy of the prediction.Based on the time series model,the LSTM neural network which can fit the nonlinear components is established in this paper.The ARIMA model and LSTM neural network are synthesized to form a hybrid model to forecast the sales volume of goods.The hybrid model can neutralize the characteristics of different models,and the LSTM neural network model can make up for the non-linearity of linear ARIMA model that can not be fitted.The mixed prediction improves the effectiveness of the prediction model,and then the two models are mixed with the principle of nonnegative weight optimal combination.Through the prediction data after the mixed model,the error between the actual value and the error value is further reduced,and the prediction effect is better than that of the single model.In automotive after-sales service sales data to establish a sales forecasting model,to improve the turnover of enterprises as well as customer satisfaction,this paper studied both the linear model based on time series,and joined the influence of nonlinear factors,by training has been able to make the model prediction accuracy sales forecast model of maximum value.