Page 1 of 10
Transactions on Engineering and Computing Sciences - Vol. 12, No. 6
Publication Date: December 25, 2024
DOI:10.14738/tecs.126.18078.
Achatbi, I. (2024). Leveraging Machine Learning Techniques to Forecast and Enhance Supplier Reliability in Supply Chain
Management. Transactions on Engineering and Computing Sciences, 12(6). 54-63.
Services for Science and Education – United Kingdom
Leveraging Machine Learning Techniques to Forecast and
Enhance Supplier Reliability in Supply Chain Management
Iman Achatbi
National School of Applied Sciences – Tangier
ABSTRACT
In today's competitive market, selecting reliable suppliers is crucial to ensure
supply chain efficiency. In fact, anticipating supplier behavior plays a vital role in
effectively managing the risk of disruptions, enabling companies to develop
proactive strategies to mitigate potential supply chain interruptions. The
complexity of supplier management has driven companies to adopt artificial
intelligence (AI) and machine learning, to enhance decision-making in the
upstream supply chain. By analyzing historical data, machine learning models help
predict risks and improve supplier reliability. These predictive capabilities allow
businesses to identify vulnerabilities early, ensuring better risk preparedness and
supply chain resilience. This article examines the application of AI to tackle supplier
selection challenges, emphasizing its role in transforming supply chains into agile,
data-driven, and predictive systems while addressing the critical need to manage
disruption risks effectively.
Keywords: Artificial intelligence, Machine learning, Supplier reliability, Logistic
Regression.
INTRODUCTION
The advent of Logistics 4.0—the integration of advanced digital technologies into supply chain
management—has revolutionized the industrial landscape. By incorporating automation, IoT
(Internet of Things), and data analytics, Logistics 4.0 has enabled companies to optimize
operations, enhance visibility, and improve decision-making processes. However, despite
these advancements, the reliability of suppliers remains a persistent challenge, particularly in
upstream logistics, where unpredictable supplier behavior can disrupt production schedules,
increase costs, and reduce customer satisfaction.
The issue of unreliable suppliers is compounded by their tendency to deliver late, fail to meet
quality standards, or provide inconsistent communication, leading to cascading effects across
the supply chain. Anticipating supplier behavior, therefore, is critical to mitigating risks,
ensuring production continuity, and maintaining competitiveness. The ability to foresee a
supplier's reliability—or lack thereof—is no longer just an operational need but a strategic
imperative in modern supply chains. In this context, the use of Artificial Intelligence (AI) in
supply chain management has emerged as a transformative solution. AI-powered tools enable
organizations to analyze large datasets, uncover hidden patterns, and make informed
predictions about supplier performance. Specifically, in upstream logistics, AI offers the
potential to evaluate supplier reliability based on historical data, enabling proactive
decisionmaking and fostering a resilient supply network.
Page 2 of 10
55
Achatbi, I. (2024). Leveraging Machine Learning Techniques to Forecast and Enhance Supplier Reliability in Supply Chain Management. Transactions
on Engineering and Computing Sciences, 12(6). 54-63.
URL: http://dx.doi.org/10.14738/tecs.126.18078
The primary objective of this article is to develop a machine learning (ML) model tailored to
assist production facilities in predicting supplier behavior. By leveraging advanced algorithms,
the proposed model aims to provide actionable insights, allowing factories to classify suppliers
as reliable or unreliable and make informed decisions about sourcing strategies. This proactive
approach not only enhances operational efficiency but also builds a foundation for sustainable
and resilient supply chain ecosystems.
By addressing these challenges, this research contributes to the growing body of knowledge
on leveraging AI in supply chain management and highlights the critical importance of supplier
reliability in achieving Logistics 4.0 goals.
LITERATURE REVIEW
The integration of artificial intelligence (AI) and machine learning (ML) into the supply chain
has gained significant attention in recent years. These technologies have demonstrated
immense potential in optimizing various aspects of supply chain management, such as demand
forecasting, inventory optimization, and route planning. However, despite the increasing focus
on AI-driven solutions, the application of these tools in supplier selection remains relatively
underexplored. According to the study conducted by (Naqvi, 2021), several studies have
addressed supplier selection using traditional methods such as Multi-Criteria Decision-Making
(MCDM) techniques, the most popular techniques in this field are: Fuzzy TOPSIS (Achatbi,
2020), Fuzzy multi objective programming (Babic, 2014), Stochastic programming (Manerba,
2018), and Mixed-integer linear programming (Sodenkamp, 2016).
There is a significant gap in utilizing AI and ML methods for supplier selection, particularly
within the context of circular supply chains. According to a review conducted by (Farshadfar,
2024), only a minority of published studies have explored supplier selection using these
advanced technologies. This gap highlights an opportunity for innovative approaches to
improve the reliability and efficiency of supplier evaluation processes. (Fallahpour, 2016)
focused on supplier selection using hybrid AI-based models and introduced innovative
methods combining 'data envelopment analysis' and 'artificial neural networks.' Additionally,
it proposed a novel genetic programming (GP) approach to enhance decision-making efficiency
in this context. (Kamalahmadi, 2016) developed a two-stage mixed-integer programming
model to create flexible sourcing strategies that mitigate supply and environmental risks while
minimizing total supply chain costs. (Tavana, 2016) proposed a Multi-Layer Perceptron (MLP)
to predict and rank supplier performance. (Choy, 2003) introduced a combined ANN-based
model designed to select and benchmark potential partners for Honeywell Consumer Products
Limited in Hong Kong. (Sawik, 2011) analyzed supplier selection and order allocation decisions
by considering factors such as the price and quality of purchased components, along with the
reliability of deliveries, in scenarios involving both local and global disruptions. (Ruiz-Torres,
2013) examined scenarios involving multiple demand points and suppliers, each characterized
by distinct costs and reliability levels. A decision-tree approach was employed to evaluate all
possible contingencies in the event of failure by one or more suppliers.
Our work seeks to address this need by developing an AI-driven model to classify suppliers
based on their reliability. By integrating multiple factors that influence supplier performance,
Page 3 of 10
56
Transactions on Engineering and Computing Sciences (TECS) Vol 12, Issue 6, December - 2024
Services for Science and Education – United Kingdom
such as delivery timelines, product quality, and after-sales service, we aim to create a
comprehensive and scalable framework. This approach not only helps in identifying reliable
suppliers but also aids in the formation of a robust database of trustworthy partners,
enhancing decision-making for procurement teams. The result is a smarter, data-driven
strategy for supplier management that aligns with the broader goals of modern supply chain
innovation.
Problem Formulation
The goal of this study is to address the challenge of classifying suppliers into reliable and
unreliable categories based on a set of criteria that describe both the supplier's characteristics
and the attributes of the products supplied. A reliable classification system is essential for
enhancing the efficiency and robustness of upstream logistics by predicting the behavior of
suppliers, including those not previously engaged.
To achieve this, the study proposes to:
1. Identify key criteria (features) that influence supplier reliability, encompassing
supplier-specific attributes (e.g., response time, after-sales service quality, financial
stability) and product-specific metrics (e.g., delivery punctuality, quality compliance).
2. Determine the weight of each criterion to reflect its relative importance in assessing
supplier reliability.
3. Develop a predictive model using the logistic regression algorithm to classify suppliers
as reliable or unreliable, providing a straightforward yet effective approach to supplier
evaluation.
Given the lack of a comprehensive real-world dataset, the study will create a synthetic dataset
to simulate a realistic environment. This dataset will include records of orders made to various
suppliers, alongside relevant supplier and product information. The generated data will serve
as a testing ground for validating the model's accuracy and effectiveness in classifying
suppliers and anticipating the behavior of new ones.
By tackling this problem, the research aims to contribute to the optimization of supplier
selection processes, reduce risks associated with unreliable suppliers, and improve overall
supply chain resilience.
PROBLEM SOLVING METHOD
Features Impacting Supplier Reliability
Features that impact supplier reliability can be categorized into different groups based on
supplier attributes and the products or services they provide. In our study, we aim to classify
the criteria impacting supplier reliability. These criteria can be grouped into three main
categories:
• Product-Related Criteria:
o Product Quality: Whether the supplied product meets the agreed-upon
standards.
o Order Accuracy: Whether the quantity delivered matches the quantity ordered.
o Unit Price: The price of the product per unit as agreed in the order.
• Delivery-Related Criteria: