FCP Filter: A Dynamic Clustering-Prediction Framework for Customer Behavior

Zhang, Y; Luo, L; Wang, Y; Wang, Z

FCP Filter: A Dynamic Clustering-Prediction Framework for Customer Behavior

Zhang, Y Luo, L Wang, Y

Wang, Z

Permalink

Publisher:: Springer Link
Publication Type:: Conference Proceeding
Citation:: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2020, 12084 LNAI, pp. 580-591
Issue Date:: 2020-01-01

Closed Access

	Filename	Description	Size
	Advances in Knowledge Discovery and Data Mining.pdf	Published version	1.33 MB	Adobe PDF	View/Open

Copyright Clearance Process

Recently Added
In Progress
Closed Access

This item is closed access and not available.

Full metadata record

Field	Value	Language
dc.contributor.author	Zhang, Y
dc.contributor.author	Luo, L
dc.contributor.author	Wang, Y https://orcid.org/0000-0002-6815-0879
dc.contributor.author	Wang, Z
dc.date	2020-05-11
dc.date.accessioned	2021-05-14T01:16:38Z
dc.date.available	2021-05-14T01:16:38Z
dc.date.issued	2020-01-01
dc.identifier.citation	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2020, 12084 LNAI, pp. 580-591
dc.identifier.isbn	9783030474256
dc.identifier.issn	0302-9743
dc.identifier.issn	1611-3349
dc.identifier.uri	http://hdl.handle.net/10453/148894
dc.description.abstract	Customer purchase behavior prediction plays an important role in modern retailing, but the performance of this task is often limited by the randomness of individual historic transaction data. In the meanwhile, Fragmentation and Coagulation Process (FCP), a stochastic partition model, has recently been proposed for identifying dynamic customer groups and modeling their purchase behavior. However, FCP is not able to forecast the purchase behavior because such a data-driven method requires transaction observations to conduct clustering. To tackle this challenge, we propose FCP filter, a clustering-prediction framework based on FCP, which can forecast purchase behavior and filter random noise of individual transaction data. In our model, FCP clusters customers into groups by their temporal interests to filter random noise of individual transaction data. Then a predictor is built on grouped data. The predicted results are also fed to FCP to adjust the parameter for prior knowledge at the next time step. Our model is superior in capturing temporal dynamics and having flexible number of groups. We conduct experiments on both synthetic and real-world datasets, demonstrating that our model is able to discover the latent group of individual customers and provides accurate predictions for dynamic purchase behavior.
dc.language	en
dc.publisher	Springer Link
dc.relation.ispartof	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
dc.relation.ispartof	Pacific-Asia Conference on Knowledge Discovery and Data Mining
dc.relation.isbasedon	10.1007/978-3-030-47426-3_45
dc.rights	info:eu-repo/semantics/closedAccess
dc.subject.classification	Artificial Intelligence & Image Processing
dc.title	FCP Filter: A Dynamic Clustering-Prediction Framework for Customer Behavior
dc.type	Conference Proceeding
utslib.citation.volume	12084 LNAI
utslib.location.activity	Singapore, Singapore
pubs.organisational-group	/University of Technology Sydney
pubs.organisational-group	/University of Technology Sydney/Faculty of Engineering and Information Technology
pubs.organisational-group	/University of Technology Sydney/Faculty of Engineering and Information Technology/A/DRsch The Data Science Institute
utslib.copyright.status	closed_access	*
pubs.consider-herdc	true
dc.date.updated	2021-05-14T01:16:37Z
pubs.finish-date	2020-05-14
pubs.publication-status	Published
pubs.start-date	2020-05-11
pubs.volume	12084 LNAI

Abstract:

Customer purchase behavior prediction plays an important role in modern retailing, but the performance of this task is often limited by the randomness of individual historic transaction data. In the meanwhile, Fragmentation and Coagulation Process (FCP), a stochastic partition model, has recently been proposed for identifying dynamic customer groups and modeling their purchase behavior. However, FCP is not able to forecast the purchase behavior because such a data-driven method requires transaction observations to conduct clustering. To tackle this challenge, we propose FCP filter, a clustering-prediction framework based on FCP, which can forecast purchase behavior and filter random noise of individual transaction data. In our model, FCP clusters customers into groups by their temporal interests to filter random noise of individual transaction data. Then a predictor is built on grouped data. The predicted results are also fed to FCP to adjust the parameter for prior knowledge at the next time step. Our model is superior in capturing temporal dynamics and having flexible number of groups. We conduct experiments on both synthetic and real-world datasets, demonstrating that our model is able to discover the latent group of individual customers and provides accurate predictions for dynamic purchase behavior.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/148894