基于机器学习的自适应预测式外呼算法研究

Research on Adaptive Predictive Outbound Algorithm Based on Machine Learning

预测式外呼是一种高效的触达客户的方式。但如何在尽可能提高坐席利用率的目标下又同时要控制住呼损率,一直是企业亟待解决的问题,尤其是在话务接通率波动较大的情况下。设计了一种基于机器学习技术的预测和自适应控制算法,能够精确地预测给定拨打速度下的坐席利用率和呼损率,能够自适应各类话务接通率来实时调整拨打速度。基于数值分析发现,在不同坐席数规模、客户接通率、外呼总量和通话时间的分布下,都比现有方法能更好地控制坐席利用率和呼损率。

Predictive outbound calling is an efficient way to reach customers. However, how to control the call loss rate while to improve the agent utilization rate has always been a problem that companies need to solve, especially when the call connection rate fluctuates greatly. This paper designed a predictive and adaptive control algorithm based on machine learning, which can accurately predict the agent utilization rate and call loss rate at a given dialing speed and adapt various types of call connection rates to adjust the dialing speed in real time. Based on the numerical analysis, it is found that under different distributions of the number of agents, the call connection rate, the total number of outgoing calls and the duration of call time, the agent utilization rate and call loss rate can be better controlled than the existing methods.