A Quality of Service (QoS) improvement system and method for transport network fault detection and QoS improvement so that a telecommunication network service provider can analyze the root cause on chronic performance issues and recommend potential solutions is disclosed. The system runs performance analysis on each AAV (mobile backhaul) or other transport networks and collects performance related metrics data. The system then selects a subset of the data related to certain key performance indicators (KPIs), such as latency, jitter, packet loss ratio, and availability. On this subset of KPI-related data, the system applies clustering techniques to identify clusters with similar performance issues. For each cluster, the system binds the AAV performance KPI data with one or more of the following site features—health, location, vendor, market, service type, etc.—to create a cluster map. The system can then generate inferences on root causes of the performance issues.