Improving network QoE and customer satisfaction has always been the primary target of network optimization and maintenance

while KPI index cannot refl ect the real network QoE

and the traditional way of analyzing customer satisfaction through feedback survey has great limitations. In this paper

the mapping relationship between KPI index and network QoE is deeply studied

and the perception weight factor is quantifi ed through big data mining and machine learning modeling. Based on this

a network QoE method based on KPI index is completed

which provides a new way for customer satisfaction analyzing and improvement.