<abstract xmlns="http://www.w3.org/1999/xhtml"><p>The traditional time series studies consider the time series as a whole while carrying on the trend detection; therefore not enough attention is paid to the stage characteristic. On the other hand, the piecewise linear fitting type methods for trend detection are lacking consideration of the possibility that the same node belongs to multiple trends. The above two methods are affected by the start position of the sequence. In this paper, the concept of overlapping trend is proposed, and the definition of milestone nodes is given on its base; these way not only the recognition of overlapping trend is realized, but also the negative influence of the starting point of sequence is effectively reduced. The experimental results show that the computational accuracy is not affected by the improved algorithm and the time cost is greatly reduced when dealing with the processing tasks on dynamic growing data sequence.</p></abstract>

The traditional time series studies consider the time series as a whole while carrying on the trend detection; therefore not enough attention is paid to the stage characteristic. On the other hand, the piecewise linear fitting type methods for trend detection are lacking consideration of the possibility that the same node belongs to multiple trends. The above two methods are affected by the start position of the sequence. In this paper, the concept of overlapping trend is proposed, and the definition of milestone nodes is given on its base; these way not only the recognition of overlapping trend is realized, but also the negative influence of the starting point of sequence is effectively reduced. The experimental results show that the computational accuracy is not affected by the improved algorithm and the time cost is greatly reduced when dealing with the processing tasks on dynamic growing data sequence.

Determination of the Starting Point in Time Series for Trend Detection Based on Overlapping Trend

Donlinks School of Economics and Management, University of Science and Technology Beijing, Beijing,

Cybernetics and Information Technologies

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

The traditional time series studies consider the time series as a whole while carrying on the trend detection; therefore not enough attention is paid to the...

Determination of the Starting Point in Time Series for Trend Detection Based on Overlapping Trend

Pubblicato online: 25 gen 2017

Pagine: 98 - 110

DOI: https://doi.org/10.1515/cait-2016-0080

Parole chiave
Overlapping trend, milestone nodes, trend detection, calculate cost

© 2016 Gao Xuedong et al., published by De Gruyter Open

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

Determination of the Starting Point in Time Series for Trend Detection Based on Overlapping Trend

Pubblicato online: 25 gen 2017

Pagine: 98 - 110

DOI: https://doi.org/10.1515/cait-2016-0080

Parole chiaveOverlapping trend, milestone nodes, trend detection, calculate cost

© 2016 Gao Xuedong et al., published by De Gruyter Open

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

Parole chiave
Overlapping trend, milestone nodes, trend detection, calculate cost