• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>Mathematical methods for information science and economics >Evolutionary Algebraic Prediction of Short Time Series
【24h】

Evolutionary Algebraic Prediction of Short Time Series

机译：短时间序列的进化代数预测

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

An overview on the new class of short-term time series forecasting methods based on the identification of skeleton algebraic sequences is given in this presentation. The concept of the rank of the sequence and the algebraic complexity of the observation are exploited to detect the base fragment of the time series and to extrapolate the model of the process into future. Evolutionary algorithms are used to remove the noise, to identify the skeleton algebraic sequence and to balance the forecast with the smoothed moving average estimate of the time series. The fitness functions exploited in the proposed forecasting technique are independent neither on the determinant of the Hankel matrix, nor on the error metrics. Numerical experiments with an artificially generated and real-world time series are used to illustrate the functionality of the proposed techniques. The proposed forecasting methods are especially effective when the time series is short and there are not always sufficient data to train evolutionary models.
机译：本演讲概述了基于骨架代数序列识别的新型短期时间序列预测方法。利用序列的等级和观测的代数复杂性的概念来检测时间序列的基本片段，并将过程模型外推到未来。进化算法用于消除噪声，识别骨架代数序列，并使预测与时间序列的平滑移动平均值估计保持平衡。所提出的预测技术中利用的适应度函数既不依赖于汉克尔矩阵的行列式，也不依赖于误差度量。用人工生成的和真实的时间序列进行的数值实验用于说明所提出技术的功能。当时间序列较短且并不总是有足够的数据来训练演化模型时，建议的预测方法特别有效。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号