Multiscale Analysis of Supply Network at Central Region in Japan

• Published in: IEEE ... International Conference on Data Mining workshops pp. 592 - 599
• Main Authors: Yi Zuo, Kajikawa, Yuya
• Format: Conference Proceeding
• Language: English; Japanese
• Published: IEEE 01.12.2016
• Subjects: Industries; Manufacturing; Measurement; Postal services; Supply chains; Water heating
• ISSN: 2375-9259
• DOI: 10.1109/ICDMW.2016.0090

Over the past decades, efficient supply chain management is recognized as a key strategic technology and resource in firms' success. Recent studies have been increasingly conducted network analysis in the fields of supply chain management and supplier-customer relationships. It is crucial to analyze interactional relationships and topological characteristics between suppliers and customers in supply networks rather than in traditionally linear supply chains. For this purpose, we involve a real-world supply network at central region in Japan in this paper and investigate different network structures of suppliercustomer relationships using network analysis techniques from three different scales. Firstly from the macro-scale, the network analysis approach provides clarity concerning network feature such as average path length. This feature can estimate the efficiency of material flow in the supply network, which also reveals the cost how passed to the final customers. Secondly from the meso-scale, we apply a network clustering algorithm to detect communities from the whole supply network, which is separated into five main supplier-customer communities. We also mark for individual communities via cross-location and crossindustry, which reveals the different features to bridge different business communities. Finally from the micro-scale, we extract core firms of each community using node degree-prior based on location-specific and industry-specific. We estimate the robustness of supply network by sequential elimination choice strategy of these hubs, which can also deliver efficient and effective support to supply network design and management.