| With the development of the call centers, more and more companies use them to expand markets. The quality of service of the call centers affects directly the benefice of a company. The abandon rate is one of the most important criteria for evaluating a call center. An accurate prediction of the abandon rate is very helpful for the managers to plan the work force and to improve the efficiency and the quality of service.Recently, Queueing theory based methods are used for predictive issues in call centers. But these methods have many theoretic hypotheses that do not usually conform to the reality of the call centers. Practice has proved that the predictive results of theses methods are not ideal for the complicate non-linear systems such as call centers. In the light of this situation, this paper makes a study of the mechanism of call centers and the existing predictive technologies;proposes an abandon rate prediction model based on BP Neural Network;makes the model validation and comparison using the actual data of a call center.The main contribution of this paper is as follows:1. Analyze and compare the existing predictive methods including: statistic based regressions, queuing theory based Erlang models, and Data Mining based time series forecast. Summe up the respective advantages and disadvantages of these methods and their practical application deficiencies.2. Considering the impact of the time series factor and the causal factor for abandon rates, taking advantage of the characteristics of artificial neural networks, propose an abandon rate prediction model based on BP neural network and give the specific modelling process.3. Design the experimental programme;illustrate the realization process of this model and predict the abandon rate with the historical data of a call center. Compare the experimental results with the results of the other two predictive methods, describe that the proposed model has a certain degree of practicality and superiority. |
PDF Full Text Request