Analysis And Research Of Sales Early Warning Model Based On Data Mining

As China's macroeconomy has entered a "new normal",many enterprises are faced with more fierce market competition.Sales risk randomness,destructive significantly increased,enterprises need to obtain early warning information as quickly and accurately as possible and then take targeted measures.Therefore,this paper puts forward the system structure of collection,processing and analysis based on mass market data,collects the national market data,including wholesale and retail data,brand information,customer basic data,etc.,completes the storage,processing and mining analysis of market data.On this basis,according to the actual demand of early warning of enterprise sales,this paper constructs a pre-warning model of enterprise sales based on massive data,conducts early warning analysis on the potential sales risks of enterprises and puts forward the implementation of targeted strategies.The specific research contents are as follows:1)Multi-channelthe market perception of information collection,With the help of market visits,terminal real-time collection,system input,etc.for large-scale enterprise data collection,cleaning,after Cleaning,processing and standardizing the collected massive data,through the actual data needs to establish different data dimensionality reduction model and the data are processed.The relational database and Hadoop technology are introduced to handle the massive basic data to realize the enterprise data architecture,data synchronization sharing and business data presentation etc.2)According to enterprise sales early warning business needs to study the supply and demand warning model,put forward the deposit and sale ratio and delivery early warning model,monitoring market product delivery and inventory risk and inventory risk,forecasting and earlywarning analysis of deposit and sale in advance,from the enterprise to the market level to control the sales warning,and on this basis to construct a sales warning knowledge base.3)Study on early warning model of enterprise sales price.Warning model is extended from the market level,put forward Holt-Winters price early warning model based on Improved Grey Wolf optimization algorithm,early warning analysis of the abnormal fluctuation in the price of enterprises in the market products and special status,and use the price early warning model on analysis results of the model of supply and demand analysis of model parameter correction and auxiliary alarm.4)Study on customer churn warning model of Enterprise sales warning model is extended from the customer level,from In the overall,ealize to analysis the sales of enterprises based on enterprise,market,customer perspective sequentially,proposed early warning model of customer churn based on improved PSO-BP neural network,analyze the enterprise customer churn by early warning,reduce customer churn,make enterprise sales to achieve sustainability,Based on the above warning model,the prediction effect of supply and demand model is corrected by using the customer churn early warning analysis result.5)On the basis of the above,build enterprise sales warning platform,starting from the actual needs of enterprises to price warning,sales ratio and delivery of customer churn warning warning,the formation of small module sales warning system applied to the early warning platform,early warning the potential state of the enterprise in real time.