Scaling forecasting algorithms using clustered modeling

Simple item page
dc.citation.epage65en_US
dc.citation.issueNumber1en_US
dc.citation.spage51en_US
dc.citation.volumeNumber24en_US
dc.contributor.authorGür, İ.en_US
dc.contributor.authorGüvercin, M.en_US
dc.contributor.authorFerhatosmanoglu, H.en_US
dc.date.accessioned2016-02-08T10:49:27Z
dc.date.available2016-02-08T10:49:27Z
dc.date.issued2015en_US
dc.departmentDepartment of Computer Engineeringen_US
dc.description.abstractResearch on forecasting has traditionally focused on building more accurate statistical models for a given time series. The models are mostly applied to limited data due to efficiency and scalability problems. However, many enterprise applications require scalable forecasting on large number of data series. For example, telecommunication companies need to forecast each of their customers’ traffic load to understand their usage behavior and to tailor targeted campaigns. Forecasting models are typically applied on aggregate data to estimate the total traffic volume for revenue estimation and resource planning. However, they cannot be easily applied to each user individually as building accurate models for large number of users would be time consuming. The problem is exacerbated when the forecasting process is continuous and the models need to be updated periodically. This paper addresses the problem of building and updating forecasting models continuously for multiple data series. We propose dynamic clustered modeling for forecasting by utilizing representative models as an analogy to cluster centers. We apply the models to each individual series through iterative nonlinear optimization. We develop two approaches: The Integrated Clustered Modeling integrates clustering and modeling simultaneously, and the Sequential Clustered Modeling applies them sequentially. Our findings indicate that modeling an individual’s behavior using its segment can be more scalable and accurate than the individual model itself. The grouped models avoid overfits and capture common motifs even on noisy data. Experimental results from a telco CRM application show the method is efficient and scalable, and also more accurate than having separate individual models.en_US
dc.description.provenanceMade available in DSpace on 2016-02-08T10:49:27Z (GMT). No. of bitstreams: 1 bilkent-research-paper.pdf: 70227 bytes, checksum: 26e812c6f5156f83f0e77b261a471b5a (MD5) Previous issue date: 2015en
dc.identifier.doi10.1007/s00778-014-0363-0en_US
dc.identifier.issn1066-8888
dc.identifier.urihttp://hdl.handle.net/11693/25714
dc.language.isoEnglishen_US
dc.publisherAssociation for Computing Machineryen_US
dc.relation.isversionofhttp://dx.doi.org/10.1007/s00778-014-0363-0en_US
dc.source.titleThe VLDB Journalen_US
dc.subjectAccuracyen_US
dc.subjectClustered modelingen_US
dc.subjectDynamic maintenanceen_US
dc.subjectScalable forecastingen_US
dc.subjectAlgorithmsen_US
dc.subjectIterative methodsen_US
dc.subjectNonlinear programmingen_US
dc.subjectTime seriesen_US
dc.subjectPerformanceen_US
dc.subjectStreaming dataen_US
dc.subjectTime series modelsen_US
dc.subjectForecastingen_US
dc.titleScaling forecasting algorithms using clustered modelingen_US
dc.typeArticleen_US

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
Scaling forecasting algorithms using clustered modeling.pdf
Size:
911.84 KB
Format:
Adobe Portable Document Format
Description:
Full printable version
Download

Collections

Scholarly Publications - Computer Engineering