Design And Implementation Of A Retail Enterprise Marketing Big Data Analysis System

Under the background of the big data era,the marketing market has been further refined,and the traditional marketing of the domestic retail industry has been unable to meet the needs of consumers.The domestic retail industry is faced with such problems as difficult to judge the marketing market situation,difficult to understand customer information and customer value,difficult to develop targeted marketing strategies.To solve these problems,a domestic traditional retail enterprise proposed to build its own precise marketing service platform.The platform mainly includes marketing big data analysis system,marketing activity management system and platform configuration management system.The marketing activity management system is responsible for the closed-loop management of the three stages of the marketing activities in the early,middle and late stages.The platform configuration management system is responsible for user,authority and access log management.The marketing big data analysis system is the core of the precision marketing service platform.It uses the big data related technology.Firstly,it realizes real-time collection of structured and unstructured data.Then convert and clean the data to load the data into the Hive data warehouse.Secondly,it uses the topic-oriented modeling method to build a data model in the Hive data warehouse.Finally,through data models and data analysis algorithms to provide accurate data for the company's operational analysis and marketing.The main work of this paper is as follows:(1)The physical topology of the system is designed from the three stages of data reporting,data service and data display.The physical architecture of the system is designed from the aspects of scalability,high availability and high security.The system architecture is designed from five levels: data source,data acquisition,data storage,programming model and data analysis.(2)Real-time collection and conversion of structured MySQL log data for multiple business systems and massive unstructured website front-end user behavior log data.(3)Analyze the business process of the enterprise and combine the method of data warehouse paradigm modeling to construct the data warehouse model based on the existing E-R model in each business,and establish a topic-oriented traffic model,sales model and RFM model,etc.(4)Based on data analysis algorithms such as transformation and subdivision,the flow and sales information of the enterprise are calculated to achieve refined operation.Based on clustering analysis and RFM model,the characteristic tag information of enterprise users is calculated to achieve accurate marketing.The retail enterprise marketing big data analysis system has been tested in function,expansibility and performance,which can meet the needs and expectations of the traditional domestic retail enterprise.The practical application in the past six months shows that the system can improve the opening rate of push links in marketing activities and finally increase the sales revenue.