LIBSVM回归算法在话务预测中的应用  

Application of LIBSVM regression algorithm in traffic prediction

作　　者：钟坛旺 林昭语 

机构地区：[1]福建省邮电规划设计院有限公司,福州350001

出　　处：《电信工程技术与标准化》2014年第9期80-83,共4页Telecom Engineering Technics and Standardization

摘　　要：话务量预测在工程规划建设中有重要的意义,但常用的预测方法总体准确率不高,而LIBSVM算法在解决回归问题上表现出优良的性能。本文以H市连续587天早忙时话务量样本作为LIBSVM模型训练,之后用60个样本点作为测试数据,验证LIBSVM模型的预测性能,并对结果进行均方误差和平均相对误差的指标评价分析。Traffic prediction has an important significance in the project, overall accuracy is not high by somecommon methods, but the LIBSVM algorithm shows good performance quality in solving the regressionproblems. This paper takes the samples as LIBSVM model training ,basing on the early busy traffi c of Hcity for 587 consecutive days, and verify the predicted performance of LIBSVM model by testing datafor following 60 samples. Finally, analyzes the results by the methods of evaluation error and averagerelative error.

关 键 词：话务量预测 LIBSVM 模型训练 样本集 

分 类 号：TN929.5[电子电信—通信与信息系统]