Research Of Massive Data Processing And Mining In Database Marketing Based On Hadoop

The application and development of the Internet not only promote the generationand development of emerging industries, deeply influence people’s life, but alsoprovide opportunities and challenges for traditional manufacture. For manufacturingenterprises, the conventional marketing ways, such as advertising, promotionalactivities has been far from meeting their needs. Under traditional marketing approach,it usually takes a long cycle and requires high cost to realize the collection ofcustomer data, manual analysis, and the formulation of appropriate marketingstrategies and methods. Even in the case of lack of data, the marketing strategies arestill blindly developed, which is difficult to achieve the desired results of theenterprise. While under database marketing, companies can easily collect andaccumulate customer information, build huge customer repository, and then apply thecloud computing technology to fast and accurate screening and analysis of massivedata, thus effectively mining customer data and maintaining relationship withcustomers. This paper proposed database marketing systems architecture based theHadoop cloud, in order to achieve massive data processing and storage, with takingthe Hongta Group database marketing system as an example. The main content of thisresearch involves:1) The current situation of the Hongta Group was analysed and demand analysisof the Hongta Group database marketing system was conducted. Based on demandanalysis, Hadoop distributed computing platform was targetedly researched andreviewed, then understanding its advantages, the structure and operation mechanism.Finally the feasibility of building Hongta Group enterprise private cloud throughHadoop was analysed.2) The data mining method commonly used in database marketing was explored.Accoring to the actual needs and the primary objective of the the Hongta Groupdatabase marketing system, promotion response model was constructed, thusimproving customer response rates of their promotional activities; then a promotion decision-making model was built to provide decision-makers with effective customerinformation, and help them determine promotional products and promotionalcustomer base; computing methods of customer lifetime value and customer loyaltywere designed, followed by customer loyalty early warning model as well as customerloyalty enhancement model.3) The collaborative work program combining Hadoop and relational databasewas explored. Then the processing method of applying MapReduce computationmodel to relational data in Hadoop Distributed platform was designed, and followingby discussion about how to use the optimal data set to select algorithm and then tobuild MapReduce Job data streams, achieving the target of universal designing atlower maintenance costs.4) In accordance with the actual situation of the Hongta Group, the overall systemdesign was given and the Hongta Group enterprise private cloud was realized in theapplication of Hadoop.