Dissertation/ Thesis
View in EDS

Implementation of the temporal fusion transformer for demand forecasting

Bibliographic Details
Title: Implementation of the temporal fusion transformer for demand forecasting
Authors: Llaquet Vélez, Santiago
Contributors: Zamora Fernandez, Sergi, Universitat de Barcelona, Accenture
Source: UPCommons. Portal del coneixement obert de la UPC
Universitat Politècnica de Catalunya (UPC)
Publisher Information: Universitat Politècnica de Catalunya, 2025.
Publication Year: 2025
Subject Terms: Classificació AMS::91 Game theory, economics, social and behavioral sciences::91B Mathematical economics, Àrees temàtiques de la UPC::Matemàtiques i estadística, Time Series Forecasting, Demand Forecasting, Multi-Horizon Forecasting, social and behavioral sciences::91B Mathematical economics, Classificació AMS::91 Game theory, economics, Transformer-based models, Deep Learning, Predicció (Estadística), Classificació AMS::68 Computer science::68T Artificial intelligence, Machine learning, Explainable AI, Supply Chain, Aprenentatge automàtic, Classificació AMS::62 Statistics::62M Inference from stochastic processes, Interpretability, Forecasting, Aprenentatge profund
Description: Time Series Forecasting plays a pivotal role across many domains, with Demand Forecasting being particularly essential for effective business planning and decision-making. While classical time series methods and Machine Learning models have achieved considerable success, they face limitations in capturing complex temporal dynamics, scaling to high-dimensional data, and offering robust interpretability. Deep Learning offers a promising alternative capable of addressing these limitations. The Temporal Fusion Transformer (TFT), a state-of-the-art Deep Learning architecture, advances multi-horizon forecasting by combining superior performance with interpretability. TFT addresses key challenges in traditional approaches through sequence encoders for local temporal processing, self-attention mechanisms for long-term dependencies, and gating mechanisms to filter irrelevant inputs, ensuring adaptability and scalability across diverse scenarios. This thesis presents an implementation of TFT for Demand Forecasting, focusing on architectural enhancements, improved performance over established methods, enhanced explainability, and practical applicability. It also evaluates TFT’s ability to address the common challenges of Transformer-based models in Time Series Forecasting. The findings assert TFT’s potential as a transformative tool for advancing Demand Forecasting methodologies and bridging the gap between state-of-the-art research and real-world business needs.
Document Type: Master thesis
File Description: application/pdf
Language: English
Access URL: https://hdl.handle.net/2117/424614
Rights: CC BY NC ND
Accession Number: edsair.dedup.wf.002..cb559f90b6fbd77f8b5812cc985d90ff
Database: OpenAIRE
FullText Text:
  Availability: 0
CustomLinks:
  – Url: https://explore.openaire.eu/search/publication?articleId=dedup_wf_002%3A%3Acb559f90b6fbd77f8b5812cc985d90ff
    Name: EDS - OpenAIRE (ns324271)
    Category: fullText
    Text: View record at OpenAIRE
Header DbId: edsair
DbLabel: OpenAIRE
An: edsair.dedup.wf.002..cb559f90b6fbd77f8b5812cc985d90ff
RelevancyScore: 887
AccessLevel: 3
PubType: Dissertation/ Thesis
PubTypeId: dissertation
PreciseRelevancyScore: 886.736389160156
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Implementation of the temporal fusion transformer for demand forecasting
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Llaquet+Vélez%2C+Santiago%22">Llaquet Vélez, Santiago</searchLink>
– Name: Author
  Label: Contributors
  Group: Au
  Data: Zamora Fernandez, Sergi<br />Universitat de Barcelona<br />Accenture
– Name: TitleSource
  Label: Source
  Group: Src
  Data: UPCommons. Portal del coneixement obert de la UPC<br />Universitat Politècnica de Catalunya (UPC)
– Name: Publisher
  Label: Publisher Information
  Group: PubInfo
  Data: Universitat Politècnica de Catalunya, 2025.
– Name: DatePubCY
  Label: Publication Year
  Group: Date
  Data: 2025
– Name: Subject
  Label: Subject Terms
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22Classificació+AMS%3A%3A91+Game+theory%2C+economics%2C+social+and+behavioral+sciences%3A%3A91B+Mathematical+economics%22">Classificació AMS::91 Game theory, economics, social and behavioral sciences::91B Mathematical economics</searchLink><br /><searchLink fieldCode="DE" term="%22Àrees+temàtiques+de+la+UPC%3A%3AMatemàtiques+i+estadística%22">Àrees temàtiques de la UPC::Matemàtiques i estadística</searchLink><br /><searchLink fieldCode="DE" term="%22Time+Series+Forecasting%22">Time Series Forecasting</searchLink><br /><searchLink fieldCode="DE" term="%22Demand+Forecasting%22">Demand Forecasting</searchLink><br /><searchLink fieldCode="DE" term="%22Multi-Horizon+Forecasting%22">Multi-Horizon Forecasting</searchLink><br /><searchLink fieldCode="DE" term="%22social+and+behavioral+sciences%3A%3A91B+Mathematical+economics%22">social and behavioral sciences::91B Mathematical economics</searchLink><br /><searchLink fieldCode="DE" term="%22Classificació+AMS%3A%3A91+Game+theory%22">Classificació AMS::91 Game theory</searchLink><br /><searchLink fieldCode="DE" term="%22economics%22">economics</searchLink><br /><searchLink fieldCode="DE" term="%22Transformer-based+models%22">Transformer-based models</searchLink><br /><searchLink fieldCode="DE" term="%22Deep+Learning%22">Deep Learning</searchLink><br /><searchLink fieldCode="DE" term="%22Predicció+%28Estadística%29%22">Predicció (Estadística)</searchLink><br /><searchLink fieldCode="DE" term="%22Classificació+AMS%3A%3A68+Computer+science%3A%3A68T+Artificial+intelligence%22">Classificació AMS::68 Computer science::68T Artificial intelligence</searchLink><br /><searchLink fieldCode="DE" term="%22Machine+learning%22">Machine learning</searchLink><br /><searchLink fieldCode="DE" term="%22Explainable+AI%22">Explainable AI</searchLink><br /><searchLink fieldCode="DE" term="%22Supply+Chain%22">Supply Chain</searchLink><br /><searchLink fieldCode="DE" term="%22Aprenentatge+automàtic%22">Aprenentatge automàtic</searchLink><br /><searchLink fieldCode="DE" term="%22Classificació+AMS%3A%3A62+Statistics%3A%3A62M+Inference+from+stochastic+processes%22">Classificació AMS::62 Statistics::62M Inference from stochastic processes</searchLink><br /><searchLink fieldCode="DE" term="%22Interpretability%22">Interpretability</searchLink><br /><searchLink fieldCode="DE" term="%22Forecasting%22">Forecasting</searchLink><br /><searchLink fieldCode="DE" term="%22Aprenentatge+profund%22">Aprenentatge profund</searchLink>
– Name: Abstract
  Label: Description
  Group: Ab
  Data: Time Series Forecasting plays a pivotal role across many domains, with Demand Forecasting being particularly essential for effective business planning and decision-making. While classical time series methods and Machine Learning models have achieved considerable success, they face limitations in capturing complex temporal dynamics, scaling to high-dimensional data, and offering robust interpretability. Deep Learning offers a promising alternative capable of addressing these limitations. The Temporal Fusion Transformer (TFT), a state-of-the-art Deep Learning architecture, advances multi-horizon forecasting by combining superior performance with interpretability. TFT addresses key challenges in traditional approaches through sequence encoders for local temporal processing, self-attention mechanisms for long-term dependencies, and gating mechanisms to filter irrelevant inputs, ensuring adaptability and scalability across diverse scenarios. This thesis presents an implementation of TFT for Demand Forecasting, focusing on architectural enhancements, improved performance over established methods, enhanced explainability, and practical applicability. It also evaluates TFT’s ability to address the common challenges of Transformer-based models in Time Series Forecasting. The findings assert TFT’s potential as a transformative tool for advancing Demand Forecasting methodologies and bridging the gap between state-of-the-art research and real-world business needs.
– Name: TypeDocument
  Label: Document Type
  Group: TypDoc
  Data: Master thesis
– Name: Format
  Label: File Description
  Group: SrcInfo
  Data: application/pdf
– Name: Language
  Label: Language
  Group: Lang
  Data: English
– Name: URL
  Label: Access URL
  Group: URL
  Data: <link linkTarget="URL" linkTerm="https://hdl.handle.net/2117/424614" linkWindow="_blank">https://hdl.handle.net/2117/424614</link>
– Name: Copyright
  Label: Rights
  Group: Cpyrght
  Data: CC BY NC ND
– Name: AN
  Label: Accession Number
  Group: ID
  Data: edsair.dedup.wf.002..cb559f90b6fbd77f8b5812cc985d90ff
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=edsair&AN=edsair.dedup.wf.002..cb559f90b6fbd77f8b5812cc985d90ff
RecordInfo BibRecord:
  BibEntity:
    Languages:
      – Text: English
    Subjects:
      – SubjectFull: Classificació AMS::91 Game theory, economics, social and behavioral sciences::91B Mathematical economics
        Type: general
      – SubjectFull: Àrees temàtiques de la UPC::Matemàtiques i estadística
        Type: general
      – SubjectFull: Time Series Forecasting
        Type: general
      – SubjectFull: Demand Forecasting
        Type: general
      – SubjectFull: Multi-Horizon Forecasting
        Type: general
      – SubjectFull: social and behavioral sciences::91B Mathematical economics
        Type: general
      – SubjectFull: Classificació AMS::91 Game theory
        Type: general
      – SubjectFull: economics
        Type: general
      – SubjectFull: Transformer-based models
        Type: general
      – SubjectFull: Deep Learning
        Type: general
      – SubjectFull: Predicció (Estadística)
        Type: general
      – SubjectFull: Classificació AMS::68 Computer science::68T Artificial intelligence
        Type: general
      – SubjectFull: Machine learning
        Type: general
      – SubjectFull: Explainable AI
        Type: general
      – SubjectFull: Supply Chain
        Type: general
      – SubjectFull: Aprenentatge automàtic
        Type: general
      – SubjectFull: Classificació AMS::62 Statistics::62M Inference from stochastic processes
        Type: general
      – SubjectFull: Interpretability
        Type: general
      – SubjectFull: Forecasting
        Type: general
      – SubjectFull: Aprenentatge profund
        Type: general
    Titles:
      – TitleFull: Implementation of the temporal fusion transformer for demand forecasting
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Llaquet Vélez, Santiago
      – PersonEntity:
          Name:
            NameFull: Zamora Fernandez, Sergi
      – PersonEntity:
          Name:
            NameFull: Universitat de Barcelona
      – PersonEntity:
          Name:
            NameFull: Accenture
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 01
              Type: published
              Y: 2025
          Identifiers:
            – Type: issn-locals
              Value: edsair
            – Type: issn-locals
              Value: edsairFT
ResultId 1