Prophet模型在话单量智能预测中的应用研究
Research on the application of Prophet model in intelligent prediction of call volume
陈勇 1刘贻凤 1王茵 1张梦宇1
作者信息
- 1. 中移信息技术有限公司,广东 深圳 518048
- 折叠
摘要
对于电信运营商而言,话单量预测在业务运营、IT系统建设规划以及系统运维方面都是至关重要的一环.传统的预测方法采用以人工预测为主,程序脚本统计为辅的方式来完成,预测的准确性受人为因素影响较大,且固定的规则无法表征受到多重因素影响的话单量的变化规律,因此,需要引入人工智能模型进行话单量的动态预测.通过研究并测试常用的人工智能预测模型,采用依据业务特征改进的Prophet模型对网间结算系统的话单量进行预测,预测效果相较于传统预测方法获得了显著提升.
Abstract
For telecommunications operators,call volume prediction is crucial in business operations,IT system con-struction planning,and system operation and maintenance.The traditional prediction method is mainly achieved through manual prediction,supplemented by program script statistics.The accuracy of predictions is greatly influ-enced by human factors,and fixed rules cannot represent the variation pattern of call volume affected by multiple fac-tors.Therefore,it is necessary to introduce artificial intelligence models for dynamic prediction of call volume.Com-monly used artificial intelligence prediction models were studied and tested.The Prophet model,improved based on business characteristics,was adopted for forecasting call volume of the inter-network settlement systems.The predic-tion effect has been significantly improved compared to traditional prediction.
关键词
话单量预测/Prophet模型/时间序列/人工智能Key words
call volume prediction/Prophet model/time series/artificial intelligence引用本文复制引用
出版年
2024
NETL
NSTL
万方数据