| With the coming of big data era, “Industries 4.0” characterized by intelligentialize, elaboration and individuation becomes possible. However, traditional industry companies still run in a production-centered mode, and a proper customer segmentation is ignored with bombarded advertisements or leaflets-based marketing strategy. Moreover, the marketing data size of large domestic enterprises has surged to TB level, and the complex data type and format make it hard to collect or utilize. In this paper, the customer information is collected by incorporating the popular social network services and the data mining techniques. Subsequently, a precision marketing information system is designed for manufacture enterprises with the idea of precision marketing, where the analysis of brand market and the monitoring and prediction of customer behavior change are realized. We take China Tobacco Yunnan Industry Co.Ltd for example in our research work. The main contributions of this paper are as follows:Firstly, we design the WeChat QR code for the product packaging of a famous and countrywide brand from a large-scale enterprise, and build the WeChat data collection subsystem for integrating the WeChat subscriptions of the enterprise brand and the marketing system. The customer information is collected when the customers scan the QR code of the product. The source of big data for the enterprises mainly includes the data interchange between the enterprises, the collaborative commerce system of industrial chain, WeChat customer information, etc. The industry-wide and multi-brand bid data can be effectively collected from different kinds of data sources.Secondly, we accomplish the big data mining on customer data. According to different value-dimension of the customers, we employ data mining technique for customer classification and clustering. In order to adapt the characteristic of large amount customer data and strong timeliness, an improved K-means algorithm is proposed for clustering analysis on customer data. The modification of the K-means algorithm is mainly indicated on two aspects. On one hand, the initial clustering center selection is optimized for data processing by the principle of error square summation, where the convergence rate of the proposed algorithm is improved, and the false clustering caused by inappropriate initial point random selection is effectively avoided. On the other hand, after the K-means algorithm finishes its run, we perform merging cluster based on the proximity principle until the cluster number meets the design requirements.Thirdly, we construct the customer segmentation rank model, and adopt Markov prediction method firstly on the marketing rank data of the customers for predictive parsing and overcome the shortage of current customers subdivided model which is deficient to perceive changes of market and customers. According to the conclusion of predictive parsing, the marketing strategies are timely corrected by marketing departments for the purpose of conforming the trend of customer change, gaining a larger share of the market as well as realizing the maximization of profits.Finally, we apply the collected customer data and the processed industrial big data to the enterprise marketing platform. As a result, the research implements the following features, including the overall process closed-loop management from customer information collection to marketing strategy formulation, high-efficient big data collection and storage, customer marketing rank prediction, brand marketing analysis, etc. According to the practical experiment for half a year, the precision marketing information system can process the data accurately and timely, where the marketing business requirements of the enterprise are fulfilled. In addition, the application of the system brings about good benefits for the enterprise and wins the recognition and praise, which provides effective solution and engineering practical experience for the other enterprises. |
PDF Full Text Request