Applied Reasearch Of Mobile Phone Brand Switch Model Based On Support Vector Machine Method

The plan of "Internet +" has promoted the combination of modern enterprises and modern technologies such as the mobile Internet,cloud computing and big data because the combination can not only help create more benefit and value,it can also effectively enhance people's perception of happiness in the 21st century.Under this background,due to the progress of related technologies and equipment,the data accumulation in different industries is rising at an unimaginable speed,so enterprises now are committed to using high-end algorithm in big data technologies to change the statue of "data rich but lack of knowledge",especially in the area of precise marketing.It is well-known that Customer is the god" is deeply ingrained in the area of precise marketing,and customer data in this area is rising rapidly due to the advancement of mobile devices.In the area of precise marketing,traditional researchers and administrators tend to use a few simple customer attributes such as gender,age,location to make classifications.Although it can excavate customer attributes to a certain extent,its accuracy is not high,and most of the mining results are too superficial.Under the background above,how to utilize the existing data to excavate customer attributes for better customer classification,thus helping with precise marketing is in great demand,because related researches can effectively improve the efficiency of the use of data and can better provide technical support for the precise marketing.Therefore,this research can provide great technical reference for accurate portraits of customers in precise marketing,and it has great theoretical significance and practical significance.This research is based on the customer data collected by mobile device and the data is about customers of X province who change their mobile phone brand in January 2016.From the perspective of consumption of last month,consumption of this month,the consumption rate,this research tries to excavate the relationship between those factors and the brand the customers will choose when they change their mobile phone brand.By using the support vector machine(SVM)in Python,this model will be used to predict customers' willingness of brand conversion,thus helping to provide more accurate customer portrait.First of all,relying on the project of "terminal brand demand in precise marketing" of a telecom company of the X province,this research collects many sheets whose primary key is customer code from the database.Then those sheets are joined to get the terminal customer data,and then the terminal data experiences data understanding,data cleaning,data conversion,etc.Secondly,after expatiating the operation principle of support vector machine and precise marketing process in detail,this research uses support vector machine(SVM)in Python to build the Mobile Phone Brand Switch Model.After several trainings,the result of the model accuracy reached 51%,so the Mobile Phone Brand Switch Model can be used to provide technical supports in precise marketing.In order to further improve the validity of the Mobile Phone Brand Switch Model,this study tries to join the four indexes of consumption rate.According to the previous steps,the accuracy improved,up to 62.63%.Finally,based on the steps of telecom precise marketing,this research puts forward some suggestions based on the Mobile Phone Brand Switch Model for precise marketing.This article embarks from the practical problems,uses the existing data at hand,devotes the classification algorithms in machine learning into the researches Mobile Phone Brand Switch Model.The results of this research can make great contributions to the marketing work.