This paper considers user level key performance indicators (KPI) evaluation for LTE network. Traditionally, cell level KPIs are obtained from the signalization data of soft/hard collection. However, the cell level KPIs could not be used to locate the communication problems within a cell. To solve this problem, we propose user level KPI evaluation methods. Two methods are proposed. User identification and positioning method is used to identify the position of grouped users; Data screening method is proposed to screen out the repetitious data and the erroneous data. The proposed methods are applied to the real LTE network. It is shown by the examples of application that the proposed methods can effectively refine the soft collection data and improve the efficiency of user level KPI evaluation.