An Experimental Evaluation of the Adaptive Sampling Method for Time Series Classification and Clustering
Muhammad Marwan Muhammad Fuad
2016
Abstract
Adaptive sampling is a dimensionality reduction technique of time series data inspired by the dynamic programming piecewise linear approximation. This dimensionality reduction technique yields a suboptimal solution of the problem of polygonal curve approximation by limiting the search space. In this paper, we conduct extensive experiments to evaluate the performance of adaptive sampling in 1-NN classification and k-means clustering tasks. The experiments we conducted show that adaptive sampling gives satisfactory results in the aforementioned tasks even for relatively high compression ratios.
DownloadPaper Citation
in Harvard Style
Fuad M. (2016). An Experimental Evaluation of the Adaptive Sampling Method for Time Series Classification and Clustering . In Proceedings of the 5th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM, ISBN 978-989-758-173-1, pages 48-54. DOI: 10.5220/0005694600480054
in Bibtex Style
@conference{icpram16,
author={Muhammad Marwan Muhammad Fuad},
title={An Experimental Evaluation of the Adaptive Sampling Method for Time Series Classification and Clustering},
booktitle={Proceedings of the 5th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM,},
year={2016},
pages={48-54},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0005694600480054},
isbn={978-989-758-173-1},
}
in EndNote Style
TY - CONF
JO - Proceedings of the 5th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM,
TI - An Experimental Evaluation of the Adaptive Sampling Method for Time Series Classification and Clustering
SN - 978-989-758-173-1
AU - Fuad M.
PY - 2016
SP - 48
EP - 54
DO - 10.5220/0005694600480054