Recently

with the scale of the high-railway network gradually expanding

the traditional network test optimization becomes increasingly arduous. In order to improve work effi ciency and reduce labor costs

the research and practice of the location network breakpoint method based on the high-railway user groups is carried out. The paper proposes a method which could identify the high-railway users based on MR big data in the view of both time and space dimensions. It cloud work out the network high-risk areas quickly and analyze automatically relying on the OTT locating

MR sampling information and the K-Measure algorithms. The research and practice have already made improvements obviously

which started in November 2017. It has gone through the stages of plan determination

function realization and verifi cation. By the end of March 2018

the proportion of error switching in some high-railway network was reduced by 17%

of which TOP 6 cells communities were reduced by 30%.