Abstract:In order to solve the problems of different professional abilities and complicated training work of the Customer Service Commissioner, through the preliminary processing and analysis of the data of more than 4000 Customer Service Commissioners, involving more than 300 types of business needs, such as business behavior, personal characteristics, quality inspection work orders, etc. in the past two years, 21 business ability evaluation dimensions are counted out from more than 100 million work orders, and a hybrid model of decision tree and neural network is established. The business capability impact factor selected by the decision tree model is applied to the LMBP neural network prediction model to get a more accurate business capability evaluation of customer service specialists clustering. In order to plan the training system of Customer Service Commissioner, through cluster analysis of the model, Customer Service Commissioner who is not good at dealing with a certain business is trained customized for the smallest business division, so as to improve the ability of Customer Service Commissioner"s business shortboard accurately and efficiently. At the same time, feedback the training results to the model analysis, improve the dimensions of business competence evaluation, and optimize the clustering results. Compared with the average business training time of customer service center, customized training based on the classification results can effectively save working time, and the training business scope is wider.