Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Applying graph neural network to SupplyGraph for supply chain network

Kihwan Han

TL;DR

This study compared performance of Multilayer Perceptions, Graph Convolution Network (GCN), and Graph Attention Network (GAT) on a demanding forecasting task while matching hyperparameters as feasible as possible and revealed that GAT performed best, followed by GCN and MLP.

Abstract

Supply chain networks describe interactions between products, manufacture facilities, storages in the context of supply and demand of the products. Supply chain data are inherently under graph structure; thus, it can be fertile ground for applications of graph neural network (GNN). Very recently, supply chain dataset, SupplyGraph, has been released to the public. Though the SupplyGraph dataset is valuable given scarcity of publicly available data, there was less clarity on description of the dataset, data quality assurance process, and hyperparameters of the selected models. Further, for generalizability of findings, it would be more convincing to present the findings by performing statistical analyses on the distribution of errors rather than showing the average value of the errors. Therefore, this study assessed the supply chain dataset, SupplyGraph, with better clarity on analyses processes, data quality assurance, machine learning (ML) model specifications. After data quality assurance procedures, this study compared performance of Multilayer Perceptions (MLP), Graph Convolution Network (GCN), and Graph Attention Network (GAT) on a demanding forecasting task while matching hyperparameters as feasible as possible. The analyses revealed that GAT performed best, followed by GCN and MLP. Those performance improvements were statistically significant at $α= 0.05$ after correction for multiple comparisons. This study also discussed several considerations in applying GNN to supply chain networks. The current study reinforces the previous study in supply chain benchmark dataset with respect to description of the dataset and methodology, so that the future research in applications of GNN to supply chain becomes more reproducible.

Applying graph neural network to SupplyGraph for supply chain network

TL;DR

This study compared performance of Multilayer Perceptions, Graph Convolution Network (GCN), and Graph Attention Network (GAT) on a demanding forecasting task while matching hyperparameters as feasible as possible and revealed that GAT performed best, followed by GCN and MLP.

Abstract

Supply chain networks describe interactions between products, manufacture facilities, storages in the context of supply and demand of the products. Supply chain data are inherently under graph structure; thus, it can be fertile ground for applications of graph neural network (GNN). Very recently, supply chain dataset, SupplyGraph, has been released to the public. Though the SupplyGraph dataset is valuable given scarcity of publicly available data, there was less clarity on description of the dataset, data quality assurance process, and hyperparameters of the selected models. Further, for generalizability of findings, it would be more convincing to present the findings by performing statistical analyses on the distribution of errors rather than showing the average value of the errors. Therefore, this study assessed the supply chain dataset, SupplyGraph, with better clarity on analyses processes, data quality assurance, machine learning (ML) model specifications. After data quality assurance procedures, this study compared performance of Multilayer Perceptions (MLP), Graph Convolution Network (GCN), and Graph Attention Network (GAT) on a demanding forecasting task while matching hyperparameters as feasible as possible. The analyses revealed that GAT performed best, followed by GCN and MLP. Those performance improvements were statistically significant at after correction for multiple comparisons. This study also discussed several considerations in applying GNN to supply chain networks. The current study reinforces the previous study in supply chain benchmark dataset with respect to description of the dataset and methodology, so that the future research in applications of GNN to supply chain becomes more reproducible.
Paper Structure (13 sections, 6 figures, 3 tables)

This paper contains 13 sections, 6 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Materials and Methods
  3. Data
  4. Use case
  5. Quality Assurance of Data
  6. Preprocessing
  7. Model Training
  8. Statistical analyses
  9. Results
  10. Discussion
  11. Acknowledgements
  12. Supplemental material
  13. Supplemental figure

Figures (6)

  • Figure 1: Supply chain graph for plant edge in a graph layout format (left) and adjacency matrix (right). In the graph layout format, the nodes were color-coded according to the product group. The rows and columns from the adjacency matrix represent presence of connection from row to column products.
  • Figure 2: The learning curves for MLP (left), GCN (center), GAT (right) models. x- and y-axis represent the number of epochs and mean-squared error loss, respectively. Loss values for the training and test dataset are shown in blue and orange color, respectively.
  • Figure 3: Plots for predicted (orange) and actual (blue) sales order over time in the training data. Each row represents sales order plots for the same product across the models. Each column represents sales order plots for the MLP (left), GCN (center), and GAT (right) models.
  • Figure 4: Supply chain graph for plant edge in a graph layout format (left) and adjacency matrix (right). In the graph layout format, the nodes were color-coded according to the product group. The rows and columns from the adjacency matrix represent presence of connection from row to column products.
  • Figure 5: Supply chain graph for plant edge in a graph layout format (left) and adjacency matrix (right). In the graph layout format, the nodes were color-coded according to the product group. The rows and columns from the adjacency matrix represent presence of connection from row to column products.
  • ...and 1 more figures