Research Of Data Mining Technology Based On Enterprise Marketing

At present,enterprises are facing the market environment change constantly,increasing in enterprise's operation instability.Enterprises must rely on complex environment inside and outside the market to make scientific decision to resist,weaken the risk.The data mining technology can excavate the new knowledge from the massive marketing data which produce from the marketing activity and the external factor.Taking these messy,unsystematic data as a starting point,we can forecast scientific business decision for enterprises through statistics,machine learning and other related theories and algorithms.we collect historical data from the enterprise management information system.Finally we construct a information platform based on data mining,to help enterprises develop fast,efficient and scientific marketing strategy,to solve the problems in the enterprise business activities.The main research contents are as follows:1.User information acquisition and mass data simplification.Design data collection model based on internal business reporting of cigarette enterprise,and also design data collection framework based on enterprise external consumer level.According to the data collected which takes a large amount,many kinds of problems exists.We shall put forward the process dimension and the cumulative drop of data based on MapReduce system Map,Reduce code.At last,we streamline the subsequent data model used by the amount of data.2.Design brand relational model.The realization of the principle of association study,by comparing the selection of Apriori brand correlation analysis,aiming at the shortcomings of the algorithm to generate large candidate items in the process of improvement.In order to produce cigarette member integral gift exchange data in the system as the application background,using the improved Apriori mining algorithm to establish the correlation model of cigarette brand,through the consumer behavior analysis of members,explores the existing relationships between brands and marketing decision for decision makers.3.Design business data mining methods.According to the influence factors of cigarette sales,we analysis the time series analysis method principle.Considering the existing time series algorithm,we proposed multi index prediction model based on the analysis of the combined forecasting of cigarette sales,and the prediction results are used for connecting cigarette production and marketing decisions.4.Proposed the platform technical level overall structure,confirmed this platform technical realization.The design of server load based on nginx is proposed.Based on the specific hardware in the process of project implementation,the weight distribution of server is carried out through stress testing,and the deployment of service level is constructed.The system functional requirements are analyzed using UML modeling.Determine the functional modules discussed in this paper,and take the functional modules of the specific application into implementation of the level software.At last,display system function.