According　to　the　ministry　of　industry　and　information　technology＇s　2018　statistical　bulletin　on　the　communications　industry,　the　net　increase　of　mobile　phone　users　reached　149　million　in　the　year,　bringing　the　total　number　to　1.57　billion.　How　to　classify　customers　according　to　their　value　and　put　forward　corresponding　business　strategies　for　different　value　customers　has　become　one　of　the　key　issues　in　the　field　of　telecommunications　research.　In　order　to　overcome　the　shortcomings　of　traditional　manual　classification,　this　paper　adopts　k-means　clustering　algorithm　and　SVD　algorithm　to　conduct　customer　clustering　in　combination　with　operator　data,　and　achieve　the　best　classification　of　customers　in　the　clustering　process,　and　analyzes　the　behavioral　characteristics　of　various　groups.　The　results　show　that　the　algorithm　can　effectively　improve　the　classification　efficiency　and　accuracy　of　telecom　users　and　reduce　the　errors　caused　by　traditional　classification.　Finally,　the　conclusion　and　further　work　are　given.