Hybrid Deep Learning with Geographic-Enterprise Anchoring for Cluster Identification in Sports Industry

Nowadays, the rapid growth of sports industry into a complex ecosystem is driven by digital transformation and enterprise diversity. Thus, the Deep Learning (DL) has emerged as a powerful tool with the increase of big data to uncover hidden patterns and enable intelligent decision making. However, the existing DL models fall short of aligning their clustered outputs with real-world industrial structures which led to limit their practical application in policymaking and regional planning. Hence, this research proposes a Hybrid DL with a Geographic Enterprise Anchoring (HDL-GEA) framework for effective sports industry cluster identification. Initially, the enterprise-level data are collected and enriched with real-world attributes, and then a Stacked Denoising Autoencoder (SDAE) is used to extract low-dimensional and noise-resistant feature representations. Furthermore, these embeddings are clustered using Deep Embedded Clustering with Guidance (DEC-G), which integrates known industry and spatial constraints. Then, validation metrics are used to evaluate cluster quality and interpretability and finally, the validated clusters are mapped for decision support in investment planning as well as regional sports development. The proposed HDL-GEA achieved better results in terms of precision (85.3%) than existing SDAE model.