改进P—SVM收视率预测方法及其应用研究 |
| |
引用本文: | 陈青,薛惠锋.改进P—SVM收视率预测方法及其应用研究[J].西安工业大学学报,2011(6):535-542. |
| |
作者姓名: | 陈青 薛惠锋 |
| |
作者单位: | 西北工业大学自动化学院,西安710072 |
| |
摘 要: | 为提高电视收视率预测水平,为相关机构提供决策支持,文中利用改进P—SVM模型对收视率数据进行分析预测.在方法设计上,为了更好解决模型参数选取等优化问题,引入证据框架,通过贝叶斯数据分析方法将样本先验知识引入模型优选过程,提出了用于收视率数据预测的改进P—SVM模型.为了测试方法的有效性,选取某电视机构收视率数据进行分析.算例分析表明,该方法较传统机器算法具有更好的预测能力,能够用于实际收视率数据分析.
|
关 键 词: | 收视率 P—SVM 证据框架 贝叶斯理论 |
Medified Potential Support Vector Machine and Its Application in Predicting Audience Rating |
| |
Affiliation: | CHEN Qing , XUE Hui-f eng (School of Automation, Northwestern Polytechnical University, Xi' an 710072, China) |
| |
Abstract: | The study aims to improve the predication of TV audience rating and to provide relevant bodies with the guidance for decision-making. A modified Potential Support Vector Machine (P-SVM) was used to predict TV andience rating. The modified P-SVM model can optimize the parameter selection by integrating the evidence framework and introducing the prior information of samples in the optimized parameter selection of the model by Bayesian method. Audience rating data of a TV organization were used to measure its effectiveness. The results show that the modified method has the advantages over the fraditional one of the better performance in predicting and analyzing the real audience rating. |
| |
Keywords: | audience rating l potential support vector machine evidence framework Bayesian theory |
本文献已被 维普 等数据库收录! |