Application Of Data Mining On Precision Marketing Of Home Broadband

The home broadband service of China Mobile has just started in the developing period.There are still lots of problems at present.It is not comprehensive enough,the original marketing method is too old,human and financial investment is too large,and the marketing effect is not good.This paper proposes to apply big data technology to the business and add data mining to production to assist in accurate marketing,which will save cost and increase income.This paper has completed the following steps: using the laboratory environment to build a Hadoop cluster as the underlying architecture;data extraction and pre-processing operations,data is divided into two parts: statistical data and detailed data;data Mining process of the statistical data by Python Machine Learning Library;evaluation formulas are formulated;Recommend the machine algorithm with the highest veracity of the model to optimize the parameters,to obtain the family broadband potential users' recommendation list with statistical data;to extend the mining system to the detailed class data,which can be mined again;to obtain potential users' recommendation list of the detailed data;to establish a user's home address recognition model;to draw a thermodynamic chart of potential users according to two recommendation lists;to get the ranking of the number of potential users according to two groups of recommendation lists.There are two outputs of this project: the recommendation list of potential users of home broadband integration package and the ranking of potential value of home broadband service in the housing estate.After optimizing,the accuracy of machine learning model is up to 77.80%,and the recommended conversion rate is expected to reach 40.0%.For production,the result of precise marketing is very ideal,but the learning model of this paper still has room for optimization.In the future,the production can be continuously optimized which can improve the precise marketing effect.It can solve the problems of difficult to discovery users,high marketing costand poor user perception of traditional family broadband marketing.This paper can also be extended to user segmentation,preference determination,career division,communication circle division,and even behavior trajectory prediction,so as to maximize the use of business resources,play the value of data,reduce costs and increase efficiency,and improve corporate profits.The innovation of this project is that I apply data mining based on big data technology into home broadband service,and the applicaion provides precise prediction that values much in our business.