摘要
专线运营缺少从售后运维数据支撑售前营销的反馈闭环,广东联通通过大数据+AI能力,进行流量建模和客户画像,帮助网络部门基于网络和业务级流量预测,指导网络精准扩容,向销售部门提供租户套餐升级评估建议和按租户的带宽进行优化的建议。运用ARIMA(趋势预测)和Boosting(训练提升)算法,基于历史流量数据对未来1～3个月数据进行预测,指导运营商对网络进行精准扩容,预测精度高达90%以上。
According to private line operation lacks feedback closed loop from after-sale operation and maintenance data to support presale marketing,Guangdong Unicom carries out traffic modeling and customer portrait through big data + AI capability,helps the network department predict network and business-level traffic,guides network accurate expansion,and provides the sale department with evaluation suggestions for upgrading tenant packages and optimization suggestions according to tenant bandwidth. Using ARIMA(trend) and Boosting training(improve) algorithm,based on historical traffic data,it forecast next 1 to 3months data to guide the operator to accurately expand the network,and the prediction accuracy is over 90%.
作者
程亚锋
刘惜吾
叶晓斌
Cheng Yafeng;Liu Xiwu;Ye Xiaobin(China Unicom Guangdong Branch,Guangzhou 510627,China)
出处
《邮电设计技术》
2018年第12期61-63,共3页
Designing Techniques of Posts and Telecommunications
关键词
流量预测
专线运营
网络精准扩容
租户套餐升级
大数据
AI
Traffic prediction
Private line operation
Accurate network expansion
Tenant package upgrade
Big data
AI
