Research And Implementation Of User Clustering Algorithm For Telecom Big Data Based On Hadoop

With the advent of the information age,the era of big data came into being.The breakthrough in the development of big data key technologies has brought forth a large number of big data innovation companies and innovative models.The development stage of the big data industry has entered the application stage from the exploration stage,and its application in important industrial fields continues to deepen.At present,China's mobile phone users have reached 1.38 billion,ranking first in the world.Therefore,carrying out data mining based on big data technology in the field of telecommunications has extraordinary significance.Grouping users is a means of marketing for telecommunications companies,scientifically discovering customer groups,and rationally allocating corporate resources to create greater profits for enterprises.Obviously,traditional technologies can no longer make effective decisions for companies.This paper combines big data technology and clustering mining algorithms to provide a new solution for telecommunication big data user grouping model.The main contents of this article include the following aspects:1,Describes the most advanced research,methods,and applications for high performance computing in large data applications.It covers emerging high-performance architectures including data-intensive applications,new efficient analysis strategies for improving data processing,and cutting-edge applications in machine learning.2,For user data with large data size,a data preprocessing method is proposed to remove the distortion data,noise,redundancy,and other invalid data from the original data.3,The traditional K-means clustering algorithm has many shortcomings,including the random generation of K value,sensitive to the initial clustering center,easy to fall into the local optimal state,etc.This paper proposes an improved K-means algorithm.Experiments show that this algorithm effectively improves the accuracy of user grouping.4,Traditional serial methods have been unable to meet the storage and processing of big data.This paper builds a fully distributed Hadoop cloud platform through simulation.On the one hand,it uses the HDFS distributed file system,and on the other hand,it combines the MapReduce programming model.The experimental results prove that the parallelization mode based on Hadoop platform improves the execution speed of the algorithm.In this paper,the clustering algorithm is applied to the large data of Hadoop to realize the user grouping,and to classify different value user groups scientifically,so as to provide enterprises with accurate marketing solutions.