中文摘要：

联络中心排班需要准确预测到达任务量的各种类型。联络中心任务量数据庞大复杂,针对多类型任务的预测方法进行了研究,实现了对联络中心大量任务数据进行准确预测类型任务的目的。首先分析了任务量数据特点,确定了工作日和休息日对任务类型的影响。工作日使用PSO优化LIBSVM模型的参数,作为弱分类器用自适应增强算法迭代训练,提出用加权投票方法融合弱分类器的思想;休息日直接使用PSO-SVM模型预测。实验结果表明,该方法的分类准确率相对于传统方法准确率有很大提高。这对联络中心的人员排班管理具有重要的作用,对其他的服务行业的任务预测分类也有重要的参考价值。

英文摘要：

All types of tasks are required to forecast accurately for shift scheduling in. contact center. According to the characteristics of tasks in. ccontact center, a multi-classifiction. method is proposed for all types of tasks and realizes the purpose of accurate classification, for a large number of tasks. First, the task characteristics is analyzed on. the task-types under the impact of working days and rest days. A PSO-SVM model with adap-tive enhancement algorithm is used as a weak classifier by using a weighted voting method in. working days, while a PSO-SVM model without adap-tive enhancement algorithm is directly classified in. rest days. Then., according to the features, a weighted majority voting method is selected to com-bine the weak classifiers into a strong classifier. The result shows that the classification, accuracy of this method is greatly improved comparing to the traditional classification, method of SVM and PSO-SVM. The method is important to scheduling management in. contact center and has a great reference value for task forecasting and classification, in. other service fields.