Clustering Time Series Data Using Kohonen Self-organizing Map (KSOM) for Classification of Sewage Treatment Plant Operations

Clustering Time Series Data Using Kohonen Self-organizing Map (KSOM) for Classification of Sewage Treatment Plant Operations

Kohonen Self Organizing Map (SOM) is an unsupervised neural network that can create a low-dimensional representation of the high-dimensional input data, allowing for visual analysis and interpretation of the clusters. In the context of time series data, SOMs can be used to identify patterns and simi...

Full description

Bibliographic Details
Published in:Lecture Notes in Networks and Systems
Main Author: Khairuddin U.; Ahmad A.; Fauzi M.F.M.; Chin K.B.; Zainudin S.F.; Aris A.M.
Format: Conference paper
Language:English
Published: Springer Science and Business Media Deutschland GmbH 2024
Online Access:https://www.scopus.com/inward/record.uri?eid=2-s2.0-85211360182&doi=10.1007%2f978-3-031-70687-5_18&partnerID=40&md5=f83f8668ef1b98d67ca048411eb7f335
id 2-s2.0-85211360182
spelling 2-s2.0-85211360182
Khairuddin U.; Ahmad A.; Fauzi M.F.M.; Chin K.B.; Zainudin S.F.; Aris A.M.
Clustering Time Series Data Using Kohonen Self-organizing Map (KSOM) for Classification of Sewage Treatment Plant Operations
2024
Lecture Notes in Networks and Systems
1133 LNNS

10.1007/978-3-031-70687-5_18
https://www.scopus.com/inward/record.uri?eid=2-s2.0-85211360182&doi=10.1007%2f978-3-031-70687-5_18&partnerID=40&md5=f83f8668ef1b98d67ca048411eb7f335
Kohonen Self Organizing Map (SOM) is an unsupervised neural network that can create a low-dimensional representation of the high-dimensional input data, allowing for visual analysis and interpretation of the clusters. In the context of time series data, SOMs can be used to identify patterns and similarities in the temporal domain and group similar time series into clusters, which can be helpful for anomaly detection and prediction tasks. Thus, this study extensively explores the KSOM algorithm to proficiently cluster data acquired from Indah Water Konsortium, a principal entity in Malaysia’s wastewater management sector. Methodically, 34,184 samples about essential wastewater treatment parameters were analyzed through a systematic data acquisition, preparation, and clustering procedure. Quantitative findings elucidated that the optimal clustering, identified with a 10 × 10 map size after 700 epochs, yielded a quantization error of 0.32 and a topological error of 0.04, producing three distinct clusters with 15.67% overlap. However, challenges emerged due to non-linear separability, complicating the delineation between the closely related clusters. The unveiled insights underscore the necessity for refined methodologies and innovative clustering techniques, propelling advancements in the precision of environmental data analysis, essential for progress in sustainable wastewater treatment processes in evolving developmental landscapes. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.
Springer Science and Business Media Deutschland GmbH
23673370
English
Conference paper
author Khairuddin U.; Ahmad A.; Fauzi M.F.M.; Chin K.B.; Zainudin S.F.; Aris A.M.
spellingShingle Khairuddin U.; Ahmad A.; Fauzi M.F.M.; Chin K.B.; Zainudin S.F.; Aris A.M.
Clustering Time Series Data Using Kohonen Self-organizing Map (KSOM) for Classification of Sewage Treatment Plant Operations
author_facet Khairuddin U.; Ahmad A.; Fauzi M.F.M.; Chin K.B.; Zainudin S.F.; Aris A.M.
author_sort Khairuddin U.; Ahmad A.; Fauzi M.F.M.; Chin K.B.; Zainudin S.F.; Aris A.M.
title Clustering Time Series Data Using Kohonen Self-organizing Map (KSOM) for Classification of Sewage Treatment Plant Operations
title_short Clustering Time Series Data Using Kohonen Self-organizing Map (KSOM) for Classification of Sewage Treatment Plant Operations
title_full Clustering Time Series Data Using Kohonen Self-organizing Map (KSOM) for Classification of Sewage Treatment Plant Operations
title_fullStr Clustering Time Series Data Using Kohonen Self-organizing Map (KSOM) for Classification of Sewage Treatment Plant Operations
title_full_unstemmed Clustering Time Series Data Using Kohonen Self-organizing Map (KSOM) for Classification of Sewage Treatment Plant Operations
title_sort Clustering Time Series Data Using Kohonen Self-organizing Map (KSOM) for Classification of Sewage Treatment Plant Operations
publishDate 2024
container_title Lecture Notes in Networks and Systems
container_volume 1133 LNNS
container_issue
doi_str_mv 10.1007/978-3-031-70687-5_18
url https://www.scopus.com/inward/record.uri?eid=2-s2.0-85211360182&doi=10.1007%2f978-3-031-70687-5_18&partnerID=40&md5=f83f8668ef1b98d67ca048411eb7f335
description Kohonen Self Organizing Map (SOM) is an unsupervised neural network that can create a low-dimensional representation of the high-dimensional input data, allowing for visual analysis and interpretation of the clusters. In the context of time series data, SOMs can be used to identify patterns and similarities in the temporal domain and group similar time series into clusters, which can be helpful for anomaly detection and prediction tasks. Thus, this study extensively explores the KSOM algorithm to proficiently cluster data acquired from Indah Water Konsortium, a principal entity in Malaysia’s wastewater management sector. Methodically, 34,184 samples about essential wastewater treatment parameters were analyzed through a systematic data acquisition, preparation, and clustering procedure. Quantitative findings elucidated that the optimal clustering, identified with a 10 × 10 map size after 700 epochs, yielded a quantization error of 0.32 and a topological error of 0.04, producing three distinct clusters with 15.67% overlap. However, challenges emerged due to non-linear separability, complicating the delineation between the closely related clusters. The unveiled insights underscore the necessity for refined methodologies and innovative clustering techniques, propelling advancements in the precision of environmental data analysis, essential for progress in sustainable wastewater treatment processes in evolving developmental landscapes. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.
publisher Springer Science and Business Media Deutschland GmbH
issn 23673370
language English
format Conference paper
accesstype
record_format scopus
collection Scopus
_version_ 1820775436769034240

Similar Items