Data Segmentation Of China Telecom Industry Under Big Data

Since 2013,4G communications technology has developed rapidly,the number of mobile phone 4G users were rapid blowout,the end of 2016 4G mobile phone users accounted for more than half of the user market,up to 770 million user groups.These massive customers need to be better managed.User segmentation has been the basis for traditional customer management systems.But the traditional telecom industry users in the sub-rough and practical and poor,can not meet the needs of customer analysis and marketing;user segmentation of academic research and the existence of complex models,the results of staggered,poor application and other issues.Leading to China's telecommunications industry on the user segmentation level is not high.4G not only represents the progress of communication technology,but also contains the user consumption structure changes.Through the Unicom 4G users to subdivide and found in the traditional three major business voice,SMS and traffic,4G user use behavior has undergone tremendous changes.3G users compared to significant changes in the level of voice consumption significantly reduced,SMS business shrinkage,the rapid growth of traffic demand for business phenomenon.At the same time for different professional attributes of the customer also user segmentation,focusing on the existing user business consumption behavior.And the standard office workers and college students of the user groups and existing 4G package matching degree analysis,and put forward marketing recommendations.Concerned about the differences in the use of professional attributes for users.While 4G technology has revolutionized the telecom industry,the development of large data technology has also brought new changes to data storage and data analysis.In the field of data mining large data analysis direction nearly two years of rapid development.In the past,the user analysis model was constrained by the platform,and the sampling method was adopted,which could not satisfy the whole and real data analysis.Traditional analysis using SPSS,SAS and other analysis software,has been unable to meet the actual needs of data analysis.Through the authority of the survey show distributed processing platform spark momentum of development,and practical strong.So,for the Unicom million level of the number of users use the spark platform for analysis,using their own Scala language programming.The K-means clustering algorithm of machine learning part is selected to construct the user segmentation model.