A Review of Subsequence Time Series Clustering

Zolhavarieh, Seyedjamal; Aghabozorgi, Saeed; Wah, Teh Ying

doi:10.1155/2014/312521

Cited by 76 publications

(38 citation statements)

References 68 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Methods include several variations of dynamic time warping [3,23,25,39], symbolic representations [29,30], and rule-based motif discovery [11,28]. There has also been work on simultaneous clustering and segmentation of time series data, which is known as time point clustering [15,49]. However, these methods generally rely on distance-based metrics, which in certain situations have been shown to yield unreliable results [24].…”

Section: Introductionmentioning

confidence: 99%

Toeplitz Inverse Covariance-Based Clustering of Multivariate Time Series Data

Hallac

Vare

Boyd

et al. 2017

Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

197

154

View full text Add to dashboard Cite

Subsequence clustering of multivariate time series is a useful tool for discovering repeated patterns in temporal data. Once these patterns have been discovered, seemingly complicated datasets can be interpreted as a temporal sequence of only a small number of states, or clusters. For example, raw sensor data from a fitness-tracking application can be expressed as a timeline of a select few actions (i.e., walking, sitting, running). However, discovering these patterns is challenging because it requires simultaneous segmentation and clustering of the time series. Furthermore, interpreting the resulting clusters is difficult, especially when the data is high-dimensional. Here we propose a new method of model-based clustering, which we call Toeplitz Inverse Covariance-based Clustering (TICC). Each cluster in the TICC method is defined by a correlation network, or Markov random field (MRF), characterizing the interdependencies between different observations in a typical subsequence of that cluster. Based on this graphical representation, TICC simultaneously segments and clusters the time series data. We solve the TICC problem through alternating minimization, using a variation of the expectation maximization (EM) algorithm. We derive closed-form solutions to efficiently solve the two resulting subproblems in a scalable way, through dynamic programming and the alternating direction method of multipliers (ADMM), respectively. We validate our approach by comparing TICC to several state-of-the-art baselines in a series of synthetic experiments, and we then demonstrate on an automobile sensor dataset how TICC can be used to learn interpretable clusters in real-world scenarios.

show abstract

Section: Introductionmentioning

confidence: 99%

Toeplitz Inverse Covariance-Based Clustering of Multivariate Time Series Data

Hallac

Vare

Boyd

et al. 2017

Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

197

154

View full text Add to dashboard Cite

show abstract

“…Os métodos de agrupamento podem ser divididos em: particionamento, métodos hierárquicos, baseados em: densidade, estruturas em grid e/ou modelos [25]. Na classificação é comum realizar pré-processamento de dados, ou seja, pode-se combinar propriedades ou usar padrões identificados na série temporal.…”

Section: Resultsunclassified

Mineração De Dados Aplicada À Séries Temporais Resultantes De Sinais Motores Da Doença De Parkinson

Folador¹,

Andrade²

2018

Anais Do v Congresso Brasileiro De Eletromiografia E Cinesiologia E X Simpósio De Engenharia Biomédica

View full text Add to dashboard Cite

Resumo: Nos últimos anos, a quantidade de dados gerados na área da saúde tem aumentado consideravelmente. Neste sentido, é necessário extrair informações destes dados, a fim de auxiliar na tomada de decisão. O trabalho aqui proposto consiste na revisão do estado atual da literatura e identificação de técnicas de mineração de dados que têm sido utilizadas na extração de informações de séries temporais, e que resultem do registro de sinais na área da engenharia biomédica e sinais motores da doença de Parkinson. Os resultados da revisão sugerem que os métodos mais utilizados na mineração de dados representados por séries temporais são: Árvore

show abstract

“…With regard the time-series clustering, different algorithms have been developed in the specific literature considering a wide set of dissimilarity or distance measures [60]. One of the most popular and field-tested similarity measures is the Dynamic Time Warping (DTW) distance, based on the optimum warping path between time-series [61].…”

Section: B Dissimilarity Measuresmentioning

confidence: 99%

Vertical Wind Profile Characterization and Identification of Patterns Based on a Shape Clustering Algorithm

et al. 2019

View full text Add to dashboard Cite

Wind power plants are becoming a generally accepted resource in the generation mix of many utilities. At the same time, the size and the power rating of individual wind turbines have increased considerably. Under these circumstances, the sector is increasingly demanding an accurate characterization of vertical wind speed profiles to estimate properly the incoming wind speed at the rotor swept area and, consequently, assess the potential for a wind power plant site. This paper describes a shape-based clustering characterization and visualization of real vertical wind speed data. The proposed solution allows us to identify the most likely vertical wind speed patterns for a specific location based on real wind speed measurements. Moreover, this clustering approach also provides characterization and classification of such vertical wind profiles. This solution is highly suitable for a large amount of data collected by remote sensing equipment, where wind speed values at different heights within the rotor swept area are available for subsequent analysis. The methodology is based on z-normalization, shape-based distance metric solution, and the Ward-hierarchical clustering method. Real vertical wind speed profile data corresponding to a Spanish wind power plant and collected by using commercial Windcube equipment during several months are used to assess the proposed characterization and clustering process, involving more than 100 000 wind speed data values. All analyses have been implemented using open-source R-software. From the results, at least four different vertical wind speed patterns are identified to characterize properly over 90% of the collected wind speed data along the day. Therefore, alternative analytical function criteria should be subsequently proposed for vertical wind speed characterization purposes. INDEX TERMS Clustering algorithms, patterns clustering, wind power generation.

show abstract

A Review of Subsequence Time Series Clustering

Cited by 76 publications

References 68 publications

Toeplitz Inverse Covariance-Based Clustering of Multivariate Time Series Data

Toeplitz Inverse Covariance-Based Clustering of Multivariate Time Series Data

Mineração De Dados Aplicada À Séries Temporais Resultantes De Sinais Motores Da Doença De Parkinson

Vertical Wind Profile Characterization and Identification of Patterns Based on a Shape Clustering Algorithm

Contact Info

Product

Resources

About