Cargando…

Time series extrinsic regression: Predicting numeric values from time series data

This paper studies time series extrinsic regression (TSER): a regression task of which the aim is to learn the relationship between a time series and a continuous scalar variable; a task closely related to time series classification (TSC), which aims to learn the relationship between a time series a...

Descripción completa

Detalles Bibliográficos
Autores principales: Tan, Chang Wei, Bergmeir, Christoph, Petitjean, François, Webb, Geoffrey I.
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Springer US 2021
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7951134/
https://www.ncbi.nlm.nih.gov/pubmed/33727888
http://dx.doi.org/10.1007/s10618-021-00745-9
_version_ 1783663593315631104
author Tan, Chang Wei
Bergmeir, Christoph
Petitjean, François
Webb, Geoffrey I.
author_facet Tan, Chang Wei
Bergmeir, Christoph
Petitjean, François
Webb, Geoffrey I.
author_sort Tan, Chang Wei
collection PubMed
description This paper studies time series extrinsic regression (TSER): a regression task of which the aim is to learn the relationship between a time series and a continuous scalar variable; a task closely related to time series classification (TSC), which aims to learn the relationship between a time series and a categorical class label. This task generalizes time series forecasting, relaxing the requirement that the value predicted be a future value of the input series or primarily depend on more recent values. In this paper, we motivate and study this task, and benchmark existing solutions and adaptations of TSC algorithms on a novel archive of 19 TSER datasets which we have assembled. Our results show that the state-of-the-art TSC algorithm Rocket, when adapted for regression, achieves the highest overall accuracy compared to adaptations of other TSC algorithms and state-of-the-art machine learning (ML) algorithms such as XGBoost, Random Forest and Support Vector Regression. More importantly, we show that much research is needed in this field to improve the accuracy of ML models. We also find evidence that further research has excellent prospects of improving upon these straightforward baselines.
format Online
Article
Text
id pubmed-7951134
institution National Center for Biotechnology Information
language English
publishDate 2021
publisher Springer US
record_format MEDLINE/PubMed
spelling pubmed-79511342021-03-12 Time series extrinsic regression: Predicting numeric values from time series data Tan, Chang Wei Bergmeir, Christoph Petitjean, François Webb, Geoffrey I. Data Min Knowl Discov Article This paper studies time series extrinsic regression (TSER): a regression task of which the aim is to learn the relationship between a time series and a continuous scalar variable; a task closely related to time series classification (TSC), which aims to learn the relationship between a time series and a categorical class label. This task generalizes time series forecasting, relaxing the requirement that the value predicted be a future value of the input series or primarily depend on more recent values. In this paper, we motivate and study this task, and benchmark existing solutions and adaptations of TSC algorithms on a novel archive of 19 TSER datasets which we have assembled. Our results show that the state-of-the-art TSC algorithm Rocket, when adapted for regression, achieves the highest overall accuracy compared to adaptations of other TSC algorithms and state-of-the-art machine learning (ML) algorithms such as XGBoost, Random Forest and Support Vector Regression. More importantly, we show that much research is needed in this field to improve the accuracy of ML models. We also find evidence that further research has excellent prospects of improving upon these straightforward baselines. Springer US 2021-03-11 2021 /pmc/articles/PMC7951134/ /pubmed/33727888 http://dx.doi.org/10.1007/s10618-021-00745-9 Text en © The Author(s), under exclusive licence to Springer Science+Business Media LLC, part of Springer Nature 2021 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic.
spellingShingle Article
Tan, Chang Wei
Bergmeir, Christoph
Petitjean, François
Webb, Geoffrey I.
Time series extrinsic regression: Predicting numeric values from time series data
title Time series extrinsic regression: Predicting numeric values from time series data
title_full Time series extrinsic regression: Predicting numeric values from time series data
title_fullStr Time series extrinsic regression: Predicting numeric values from time series data
title_full_unstemmed Time series extrinsic regression: Predicting numeric values from time series data
title_short Time series extrinsic regression: Predicting numeric values from time series data
title_sort time series extrinsic regression: predicting numeric values from time series data
topic Article
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7951134/
https://www.ncbi.nlm.nih.gov/pubmed/33727888
http://dx.doi.org/10.1007/s10618-021-00745-9
work_keys_str_mv AT tanchangwei timeseriesextrinsicregressionpredictingnumericvaluesfromtimeseriesdata
AT bergmeirchristoph timeseriesextrinsicregressionpredictingnumericvaluesfromtimeseriesdata
AT petitjeanfrancois timeseriesextrinsicregressionpredictingnumericvaluesfromtimeseriesdata
AT webbgeoffreyi timeseriesextrinsicregressionpredictingnumericvaluesfromtimeseriesdata