The Research And Application Of The Data Mining In Communication Network Optimization

With the development of economy and technology,network communication is becoming one of the important tools of national life,which is needed to be stability and security anytime.Network optimization is a vital method to complete this task and the premise is to have a clear understanding of the basic operation of the current network.Aiming at the low efficiency of the traditional analysis method and the characteristics of huge network data,the application of big data mining technology in network analysis is put forward.Firstly,we have a treatment to the data and remove the abnormal cells,then make a clustering for the network data.Finally,we analyze the data of each kind of cell and propose a network optimization scheme according to the operation of the network.An improved LOF algorithm is used to detect abnormal cells.This algorithm can determine the number of abnormal point by the density distribution of the network data.The experiments show that the algorithm has high accuracy and low false alarm rate,and overcomes the shortcoming that the LOF algorithm must know the number of outliers.In order to improve the accuracy and stability of the clustering algorithm,an improved K-means clustering algorithm has been presented.The traditional K-means algorithm is sensitive to initial cluster centers and the number of cluster.But the improved algorithm can solve the two defects through density measurement and distance measurement and average maximum similarity index between classes.Results of Experiments show that the improved K-means clustering algorithm can effectively boost clustering accuracy and stability and reduce the clustering time and error.Finally,the network characteristics of each cell are analyzed.We analyze the relationship between the connection device and the network utilization rate and network disconnection.For any kind of network dropped,We can get the accessibility margin of the current network and propose a network optimization scheme to avoid network overload.