TIME SERIESFORECASTINGWITHTRANSFORMER AND INFORMERMODELS

Please use this identifier to cite or link to this item: http://localhost:8081/jspui/handle/123456789/18851

Title:	TIME SERIESFORECASTINGWITHTRANSFORMER AND INFORMERMODELS
Authors:	Konduri, Issac Babu
Issue Date:	Jun-2024
Publisher:	IIT, Roorkee
Abstract:	Timeseriesforecastingwiththehelpofadvancedmachinelearninganddeeplearn- ing models,specificallyTransformersandtheInformer,isappliedtothreediverse datasets fromtheMonashTimeSeriesRepository.Thepedestriancountsdatasetin- cludes hourlycountsofpedestriansinMelbournebeginninginMay2009,asrecorded by 66sensorsinthecity,withdataavailablethroughApril30,2020.Thesolarenergy dataset encompassessolarpowerproductionrecordsfrom137photovoltaicplantsin Alabama, sampledevery10minutesthroughout2006,capturinghigh-frequencyvari- ations insolarenergyoutput.Theelectricitydemanddatasetcontainsfivetimeseries corresponding tothehalf-hourlypowerconsumptionoffiveAustralianstates:South Australia, Queensland,Tasmania,Victoria,andNewSouthWales.Emphasizingtheim- portance ofaccurateforecasting,theresearchleveragestheTransformerandInformer model, designedforefficienthandlingoflong-sequencetimeseriesdata.Thestudy demonstrates substantialimprovementsinforecastingaccuracycomparedtotraditional models. Thesefindingshighlightthepotentialofadvanceddeeplearningmodelsto enhance decision-makingprocessesinenergyplanning,urbanmanagement,andinfras- tructure developmentbyprovidingreliableandcomprehensiveforecasts.
URI:	http://localhost:8081/jspui/handle/123456789/18851
Research Supervisor/ Guide:	Pillai, G.N.
metadata.dc.type:	Dissertations
Appears in Collections:	MASTERS' THESES (MFSDS & AI)

Files in This Item:

File	Description	Size	Format
22565008_KONDURI ISSAC BABU.pdf		4.06 MB	Adobe PDF	View/Open