Power customer appeal service risk assessment method and system based on unsupervised learning

Abstract

The invention discloses an unsupervised learning-based power customer appeal service risk assessment method and system. The method comprises the following steps of: obtaining multi-channel power customer appeal data and risk factors contained in the multi-channel power customer appeal data; carrying out feature clustering on the multi-channel power customer demand data by adopting a K-Means clustering algorithm which introduces a center limit theorem and optimizes an initial clustering center, and classifying the data with the same risk factor into one cluster; according to a clustering result, respectively using two models to calculate the weight of each risk factor, obtaining the current power customer evaluation of each risk factor, and carrying out weighted calculation on the power customer evaluation through a comprehensive weight method to obtain a current overall risk value; and carrying out risk grade division, and determining a current risk grade based on the calculated total risk value. By establishing a standardized data processing flow, the accuracy and reliability of power customer appeal service risk assessment are remarkably improved, and powerful support is provided for continuous improvement of power service quality.

Images