Profiling the applicability of supervised machine learning in supply chain information systems: An agglomerative clustering approach

Gamoura, S. C. (2023) Profiling the applicability of supervised machine learning in supply chain information systems: An agglomerative clustering approach. Dans CIGI Qualita MOSIM 2023, Trois-Rivières, Québec, Canada DOI 10.60662/95wz-x073.

Prévisualisation

Texte CIGI_QUALITA_MOSIM_2023_paper_6804.pdf Télécharger (964kB) | Prévisualisation

Résumé

In recent years, Predictive Data Analytics (PDA) integrated with Supply Chain Information Systems (SCIS) has been the focus of considerable work to enable companies to make better decisions and remain competitive. Supervised Machine Learning (SML) approaches are becoming the main lever to smooth and facilitate this integration in this challenging environment. These approaches lead academics and industry to abandon procedural development and begin to think about undertaking them inductively by learning from the input data. However, SCISs have specific considerations that strongly affect the effectiveness of these learning models. Indeed, practitioners do not have some "recipe" for choosing a specific SML algorithm for a given problem in SCIS. They have to go through tedious phases and often depend on IT providers to do so. For this reason, the applicability of SML in SCIS is today an emerging challenge for scientists and practitioners, while the scientific literature is still in its early stages. This paper attempts to fill this gap by proposing a novel profiling approach for the applicability of SML in SCIS, including a comprehensive dual taxonomy with a Hierarchical Agglomerative Clustering algorithm (HAC). The profiling approach can help researchers and industrialists in the early selection of algorithms for their integration projects and thus avoid failure rates and expensive investments.

Type de document:	Document issu d'une conférence ou d'un atelier (NON SPÉCIFIÉ)
Mots-clés libres:	Artificial Intelligence Supervised Machine Learning Supply Chain Information System Profiling Applicability
Date de dépôt:	17 août 2023 15:29
Dernière modification:	11 sept. 2023 19:18
URI:	https://collection-numerique.uqtr.ca/id/eprint/2076

Actions (administrateurs uniquement)

Éditer la notice