基于双模态Transformer模型的话务量预测

Call Volume Prediction Based on Bimodal Transformer Model

为降低客户服务中心电话的等待率,提升服务质量。针对现有算法不能实现中长期话务量预测的问题,提出了一种基于双模态Transformer模型的话务量预测方法。首先采集并预处理某运营商真实的话务量数据,通过双模态特征融合构造出有益特征,最后采用多种模型进行话务量预测以及多种衡量指标对预测结果进行分析。结果表明:与其他算法比较,Transformer模型性能较好,对运营商资源的合理配置具有较高的指导意义,同时更易获得客户较高的满意度和忠诚度。

In order to reduce the waiting rate for calls from customer service centers and improve service quality,a method for call volume prediction based on dual-modal Transformer model is proposed.This is in response to the issue that existing algorithms cannot achieve medium and long-term call volume prediction.Firstly,the real call volume data from a certain ISP is collected and preprocessed;then,beneficial features are constructed through bimodal feature fusion;finally,multiple models are used for call volume prediction.Additionally,a variety of measurement indicators are used to analyze the prediction results.The results show that compared with other algorithms,the Transformer model performs better,providing significant guidance for the rational allocation of ISP resources.It also makes it easier to achieve higher customer satisfaction and loyalty.