Extraction of business relationships in supply networks using statistical learning theory

Yi Zuo, Yuya Kajikawa, Junichiro MoriHeliyon 2016 · DOI

TL;DR

This paper introduces a novel approach to predicting customer-supplier relationships in supply networks by combining traditional enterprise attributes with network centrality measures derived from social network analysis (SNA). By leveraging Support Vector Machines (SVMs), the study demonstrates that this hybrid approach significantly improves prediction accuracy, offering a more nuanced understanding of business relationships beyond simple attribute-based models.

What Problem Does This Paper Address?

Traditional supply chain management often models relationships linearly or dyadically, which fails to capture the complex, cyclical nature of real-world supply networks. This simplification hinders effective supply chain design and the identification of potential business partners. While Social Network Analysis (SNA) offers a more holistic view, its application to empirical supply network data and its integration with machine learning for predictive purposes have been limited. This paper aims to bridge this gap by developing and validating an SNA-based machine learning model that can predict customer-supplier relationships more accurately, especially for smaller enterprises.

Methods

The core of this research lies in applying a machine learning approach, specifically Support Vector Machines (SVMs), to predict customer-supplier relationships. The novelty comes from the features used:

  • Enterprise Attributes (EAs): These are traditional data points about each company, such as capital, founding date, number of employees, location, industry, sales, and profit.
  • Network Centralities (NCs): These are derived from Social Network Analysis (SNA) and quantify an enterprise's importance and influence within the supply network. The paper focuses on:
  • Degree Centrality: The number of direct connections a node has.
  • Closeness Centrality: How quickly a node can reach all other nodes.
  • Betweenness Centrality: How often a node lies on the shortest path between other nodes.

The methodology involves:

  1. Constructing the Supply Network: Using transaction data to build a network where enterprises are nodes and relationships are links.
  2. Community Detection: Employing the Newman method (a fast modularity maximization algorithm) to identify densely connected groups (communities) within the larger network.
  3. Calculating Network Centralities: Determining degree, closeness, and betweenness for each enterprise within its community.
  4. SVM Model Training: Using a combination of EAs and NCs as input features to train an SVM model to predict the existence of a customer-supplier relationship. The paper explores different kernels (linear, polynomial, RBF) and hyperparameters to find the optimal model.

Datasets

The study utilizes a comprehensive dataset of 182,538 enterprises and 598,721 transactions from central Japan, provided by Tokyo Shoko Research Limited. This dataset includes detailed enterprise attributes and transaction relationship data, forming the basis for constructing the supply network. The paper does not introduce a new benchmark dataset but rather uses this real-world data for its analysis.

Key Insights and Ideas

  • Beyond Attributes: Simply relying on enterprise attributes is insufficient for accurately predicting complex business relationships. Network structure and an enterprise's position within it are crucial.
  • The Power of Centrality: Network centrality measures (closeness and betweenness) significantly enhance predictive accuracy when combined with enterprise attributes. This highlights the importance of an enterprise's role in information flow and mediation within the network.
  • Community Matters: Analyzing supply networks at the community level, rather than as a single monolithic entity, leads to more accurate predictions. Sub-communities exhibit denser connections, making network centralities even more influential.
  • SMEs Benefit Most: Small and Medium-sized Enterprises (SMEs), often at a disadvantage in finding business partners using attribute-based methods, see substantial improvements in predictive accuracy when network centralities are incorporated. This is particularly true for SME-LE and SME-SME relationships.

Key Results

  • Overall Improvement: Combining Enterprise Attributes (EA) and Network Centralities (NC) achieved an F-value of approximately 82%, a significant improvement over using EA alone (77%) or NC alone (70%).
  • Community-Level Gains: When analyzing sub-communities, the predictive accuracy improvement from combining EA and NC was even more pronounced, with increases of 7.18% to 9.39% compared to using EA alone.
  • SME-LE and SME-SME: For these relationship types, the combined EA & NC approach showed the most dramatic improvements, particularly in F-value, suggesting it's highly effective for SMEs seeking partners. For instance, the F-value for SME-LE improved by 17.90% compared to using EA alone.
  • SVM Performance: The RBF kernel with specific hyperparameters (σ=1.0, C=1.0) yielded the best predictive accuracy among the tested SVM configurations.

Limitations and Future Work

  • Parameter Optimization: While the proposed method outperforms others, its predictive performance can still be optimized by searching for the ideal combination of parameters.
  • New Variable Generation: The authors plan to explore new variables, such as text information from company websites and news media, to gain a more nuanced perspective on actual business conditions.
  • Customized Mapping: Developing customized mapping approaches for these new text-based variables is also a future direction.
  • Generalizability: While the study focuses on a Japanese supply network, further validation on diverse datasets would be beneficial.

Summary updated

Abstract

Supply chain management represents one of the most important scientific streams of operations research. The supply of energy, materials, products, and services involves millions of transactions conducted among national and local business enterprises. To deliver efficient and effective support for supply chain design and management, structural analyses and predictive models of customer-supplier relationships are expected to clarify current enterprise business conditions and to help enterprises identify innovative business partners for future success. This article presents the outcomes of a recent structural investigation concerning a supply network in the central area of Japan. We investigated the effectiveness of statistical learning theory to express the individual differences of a supply chain of enterprises within a certain business community using social network analysis. In the experiments, we employ support vector machine to train a customer-supplier relationship model on one of the main communities extracted from a supply network in the central area of Japan. The prediction results reveal an F-value of approximately 70% when the model is built by using network-based features, and an F-value of approximately 77% when the model is built by using attribute-based features. When we build the model based on both, F-values are improved to approximately 82%. The results of this research can help to dispel the implicit design space concerning customer-supplier relationships, which can be explored and refined from detailed topological information provided by network structures rather than from traditional and attribute-related enterprise profiles. We also investigate and discuss differences in the predictive accuracy of the model for different sizes of enterprises and types of business communities.
BibTeX
@inproceedings{zuo2016extraction,
  author = {Yi Zuo and Yuya Kajikawa and Junichiro Mori},
  title = {{Extraction of business relationships in supply networks using statistical learning theory}},
  booktitle = {Heliyon},
  year = {2016},
  doi = {10.1016/j.heliyon.2016.e00123},
}
Figure