A Machine Learning-based decision support system for Fast-Moving Consumer Goods vendor selection

Lists

Bukowska, Anna and Ressin, Malte ORCID: https://orcid.org/0000-0002-8411-6793 (2024) A Machine Learning-based decision support system for Fast-Moving Consumer Goods vendor selection. In: BAM2024, 2 Sept - 6 Sept 2024, Nottingham, UK. (Submitted)

There is a more recent version of this item available.

[thumbnail of FMCG AI supply chain BAM submission.pdf]

Preview

PDF
FMCG AI supply chain BAM submission.pdf - Accepted Version
Available under License Creative Commons Attribution.
Download (433kB) | Preview

Abstract

In a competitive business environment, procurement is a profit-contributing activity, particularly for fast-moving consumer goods where delays and quality issues can result in missed sales. Classically, the main vendor selection factor is price.
This study conducts vendor classification based on performance data to reduce supply chain risk by enabling informed procurement decisions. Employed performance characteristics are product quality, delivery time, communication, reliability, and geographical distance.
Three different prototype classifiers are tested: Random Forest, K-Nearest Neighbour, and Naïve Bayes. Performance metrics are accuracy, precision, recall, specificity, and f-1 score.
Using an artificial dataset, Random Forest shows the best performance results (accuracy 0.97, f1-score, 0.86), followed by Naïve Bayes (accuracy 0.91, f1-score 0.80) and K-Nearest Neighbours (accuracy 0.88, f1-score 0.78). Random Forest also exceeds at detecting features with the most impact in the data. Results provide a first step towards the implementation of ML-based vendor classification by indicating and benchmarking suitable algorithms.

Item Type:	Conference or Workshop Item (Paper)
Keywords:	Supply-chain management, machine learning, vendor selection, artificial intelligence
Subjects:	Computing > Intelligent systems Computing
Depositing User:	Malte Reßin
Date Deposited:	29 Apr 2024 08:59
Last Modified:	04 Nov 2024 11:22
URI:	https://repository.uwl.ac.uk/id/eprint/11566