时间序列的分频与建模

发布时间：2018-04-22 17:37

本文选题：时间序列 + 频率分量　；参考：《统计与决策》2010年09期

【摘要】：文章采用分频方式分析时间序列并为时间序列建模,首先对时间序列做离散余弦变换,用低频变换系数重构出时间序列的低频分量,而剩余的高频部分则采用时滞自相关分析方法确定模型结构。针对股票多年日收盘价所作仿真试验证明,该时间序列建模方法是有效的,模型比较好地刻画了时间序列的变化规律。
[Abstract]:In this paper, the time series is analyzed by the frequency division method and the time series is modeled. First, the discrete cosine transformation is made to the time series, the low frequency component of the time series is reconstructed by the low frequency transform coefficient, while the remaining high-frequency part is used to determine the model structure by the time-delay autocorrelation analysis method. Obviously, the time series modeling method is effective, and the model depicts the change law of time series better.

【作者单位】：上海理工大学管理学院;
【基金】：上海市教育委员会科研资助项目(07ZZ94) 上海市重点学科资助项目(S30501)
【分类号】：F830;F224

【相似文献】