Study On Market Delivery Strategy Based On Massive Data

Currently in the market delivery strategy of individual subjective judgments overweight,need to increase market factors,provide scientific decision support.Digging out valuable information from historical data and forecasting the market demand more accurately,so as to formulate different delivery strategies,determine the strategic orientation and realize the "precise strike" of product launches,then we can grasp the initiative in the market,strive for more profits for the enterprise.This article has completed the massive market data collection and processing analysis,cluster analysis of many retail customers on the market to identify the characteristics of different retail customers,the establishment of market delivery strategy model for different characteristics of personalized product resources,and adjusting strategy according to other factors.Finally,through application verification,the marketing delivery strategy model proposed in this paper can well fit the historical sales data and predict the sales volume,providing technical support for effective market delivery.The specific research contents are as follows:1)Massive data collection and processing.Through the terminal system filing,market visits and terminal acquisition and other means to complete the acquisition of massive business data and data for the pre-treatment of the problematic data,and then massive data is processed synchronously,the use of HDFS storage data,MapReduce model distributed processing of data,Hive builds the data warehouse and Hbase to process and query real-time data,and finally use Sqoop tool to exchange data between Hadoop and relational database to meet the business needs.2)The choice of the target retailer classification.Market-oriented enterprise product resources put the ultimate goal is to calculate the precise amount of personalized delivery based on sales volume,inventory,storage ratio and other historical information of retail customers.Therefore,under the background of massive data,clustering mining for retailers is of great significance to guide the market delivery.In this paper,an improved CURE algorithm based on MapReduce parallel computing(introducing Mahalanobis distance to measure inter-cluster similarity)is used to realize the clustering analysis of retailers under big data,and the feasibility of CURE algorithm is verified by experiments.3)Establish the market delivery strategy model.According to the characteristics of different types of retail customers personalized product resources.Product resources in the market volume is an important factor in regulating the market progress,and market delivery forecast accuracy will determine the accuracy of the amount of delivery.According to the different types of retailers,the sales volume of certain products in the region are respectively analyzed by regression analysis and ARMA time series to predict the sales volume,stock and stock-to-sales ratio.Finally,a model of market delivery strategy is established and adjusting strategy according to other factors.In the end,it is verified that the adopted market delivery strategy model fits the historical data well and predicts the delivery volume,which provides an effective mathematical model for market delivery.4)System implementation.Set up a corporate marketing strategy system,the model used in the actual delivery of the enterprise.Application results show that the process of enterprise product delivery process more accurate,and thus make the product sales volume,sales amount,occupancy rate of production have significantly improved,and improve the sales profits of enterprises.