A Machine Learning Approach to Enhancing Supplier Quality Risk Assessment
(2025)Department of Automatic Control
- Abstract
- The possibilities of artificial intelligence (AI) have become evident in recent years, with new applications manifesting the potential. The field of supplier quality management is no exception, as AI shows promise of transforming the field. AI can boost efficiency, increase transparency while also removing human biases, and ensuring one coherent way of working across the organization. Understanding how AI can be adopted in supplier quality management can help companies improve operations. While some studies about theoretical implications have been carried out, the literature focusing on implementation is limited.
The main objective of this thesis was to study supplier quality management through an AI lens. Two concrete use cases were... (More) - The possibilities of artificial intelligence (AI) have become evident in recent years, with new applications manifesting the potential. The field of supplier quality management is no exception, as AI shows promise of transforming the field. AI can boost efficiency, increase transparency while also removing human biases, and ensuring one coherent way of working across the organization. Understanding how AI can be adopted in supplier quality management can help companies improve operations. While some studies about theoretical implications have been carried out, the literature focusing on implementation is limited.
The main objective of this thesis was to study supplier quality management through an AI lens. Two concrete use cases were identified: (1) to automate the current supplier quality risk classification, and (2) to forecast defective parts per million. Through actual implementation, the extended objective became to identify activities and infrastructure necessary for working in a data-driven manner. Internal data was gathered and preprocessed, constructing a pipeline that enabled fragmented data of various formats to be compiled into a dataset usable for modeling.
The analysis resulted in a score of 76.3% in the classification task, revealing that features measuring business risk are given much weight in the current way of working. Forecasting, however, showed little success due to extreme variance in the data. Furthermore, the factor limiting performance is the availability and quality of the data, rather than the machine learning (ML) models. The implications of the results obtained are that: (1) the potential of ML is reaffirmed, and (2) the insights generated during the thesis are of as much importance as the resulting models. They shed light on potential improvements while highlighting activities necessary for adopting ML technologies. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9205630
- author
- Gustafsson, Tobias and Hansson, William
- supervisor
- organization
- year
- 2025
- type
- H3 - Professional qualifications (4 Years - )
- subject
- report number
- TFRT-6274
- other publication id
- 0280-5316
- language
- English
- id
- 9205630
- date added to LUP
- 2025-08-08 15:12:46
- date last changed
- 2025-08-08 15:12:46
@misc{9205630,
abstract = {{The possibilities of artificial intelligence (AI) have become evident in recent years, with new applications manifesting the potential. The field of supplier quality management is no exception, as AI shows promise of transforming the field. AI can boost efficiency, increase transparency while also removing human biases, and ensuring one coherent way of working across the organization. Understanding how AI can be adopted in supplier quality management can help companies improve operations. While some studies about theoretical implications have been carried out, the literature focusing on implementation is limited.
The main objective of this thesis was to study supplier quality management through an AI lens. Two concrete use cases were identified: (1) to automate the current supplier quality risk classification, and (2) to forecast defective parts per million. Through actual implementation, the extended objective became to identify activities and infrastructure necessary for working in a data-driven manner. Internal data was gathered and preprocessed, constructing a pipeline that enabled fragmented data of various formats to be compiled into a dataset usable for modeling.
The analysis resulted in a score of 76.3% in the classification task, revealing that features measuring business risk are given much weight in the current way of working. Forecasting, however, showed little success due to extreme variance in the data. Furthermore, the factor limiting performance is the availability and quality of the data, rather than the machine learning (ML) models. The implications of the results obtained are that: (1) the potential of ML is reaffirmed, and (2) the insights generated during the thesis are of as much importance as the resulting models. They shed light on potential improvements while highlighting activities necessary for adopting ML technologies.}},
author = {{Gustafsson, Tobias and Hansson, William}},
language = {{eng}},
note = {{Student Paper}},
title = {{A Machine Learning Approach to Enhancing Supplier Quality Risk Assessment}},
year = {{2025}},
}