Optimization Model for Bidding and Procurement Packaging Based on Improved DBSCAN Algorithm

In addressing the packaging issue in tender procurement, this study introduces an enhanced DBSCAN clustering algorithm that incorporates an edge-betweenness-based edge-breaking approach. Utilizing historical tender procurement packaging data, a distance matrix among materials is constructed. The traditional DBSCAN clustering algorithm’s density-connectivity characteristic poses challenges to the granularity of clustering results. To address this, an optimization method involving edge-betweenness edge-breaking is devised. By computing the edge-betweenness of connections between data points, edges crucial to the network structure are identified and subsequently removed. This process disrupts the internal connections of “large clusters,” causing them to fragment into smaller, more refined clusters. Consequently, the accuracy and resolution of procurement packaging clustering results are enhanced. The proposed method is employed in data experiments involving items requiring tender procurement packaging and is compared with alternative clustering packaging algorithms. The results demonstrate that by optimizing the clustering structure and elevating the silhouette coefficient, this method significantly improves the accuracy and robustness of packaging results.