Research On New Retail Platform Procurement Forecast And Inventory Reserve Warning Plan

The concept of “new retail” was proposed by Ma Yun in the second half of 2016,that is,a new mode of retail in which the company relies on the Internet to transform the production,circulation and sales process of goods and to integrate online service,offline experience and modern logistics by means of advanced technologies such as big data.The purpose of the new retail is to improve the efficiency.The traditional retail platform has high sales cost,high resource consumption,high storage cost,while the integration mode of online,offline and logistics represents the direction of future development of retail platform.The procurement of new retail is particularly important.It can neither be in short supply nor oversupply.Therefore,the forecast of procurement is quite significant.When it is in short supply,to prevent customer loss,an inventory reserve warning plan should be formed to enhance corporate profits.In 2017,the “Automobile Sales Management Measures” introduced the elimination of mandatory brand authorization and encouraged the diversification of automobile sales models.Driven by the policy,the new retail of automobiles has developed rapidly.As one of the emerging industries of new retail,the automotive industry has already appeared in the market and drew a high degree of attention.This paper takes a new retail platform of a car as an example.Using the internet as a carrier,a new retail platform for a car is linked online and offline through financial leasing.The new retail platform of the car reduces the logistics time of vehicles by establishing offline warehouses,storing and purchasing vehicles,and directing them from the warehouse to consumers.Different from the procurement of other products,cars have long transportation time,high cost and high storage cost.To improve efficiency,the most important thing is to reduce inventory space,reduce inventory and increase turnover.In addition,the launch rate of new models by major automakers is fast and the change in car prices is rapid,while the models that have been purchased in advance to form inventory are limited.In order to prevent the loss of potential customers,an alternative plan should be provided to users who have a strong desire to purchase but whose intentioned models have insufficient stocks,forming an early warning.If consumers did the “loan qualification” test in the new retail platform APP,it means that they have a strong desire to buy a car,but they have not placed a purchase order.At this time,the user can be classified as a “potential user”.If there are a large number of potential users of an insufficiently stoked model,then this model should be evaluated and included in the procurement plan;if the consumers who intend to use this model are a minority,individualized demand of different consumers should be treated differently.In order to avoid excessive telephone inquiries and harassment,priority should be given to using models to find similar in-stock vehicle models that consumers can afford.This paper takes a new retail platform of a car as the research subject and the overall goal is to avoid losses and improve corporate income.First,procurement forecasts are made based on the supply side.For the models being sold,historical sales can directly provide reference for the quantity of purchases;for the new models,this paper compares the sales of the models based on different performances of RandomForest,GBRT and XGBoost algorithms,and finally selects the GBRT model as the optimal model.The output is the importance of the feature which can provide a decision on the quantity of purchases.Additionally,an inventory reserve warning plan is given based on the consumer situation.Considerations on both price and performance should be given when providing consumers with an alternative to the intended model.In terms of price,comparing the RandomForest,GBRT and XGBoost models,the car price is predicted based on the consumer's own attribute information entered in the test “Loan Qualification”.Finally,the GBRT model is selected as the optimal model,and the importance of the feature is the output.In terms of vehicle performance,using K-Prototypes algorithm to cluster all models in the past five years,in-stock models that meet the consumer's required performance attributes and are in the same category of the intended vehicle models are selected and recommended to users.The two aspects mentioned above are combined to maximize retention of customers,prevent potential customers loss,and provide reliable support for decision-making.