Account-Based Marketing in the Railway Sector: An Optimization Approach with K-means 9++w and Linear Programming

This article proposes an optimized approach for account-based marketing (ABM) in the railway sector, utilizing a variant of the K-means algorithm, dubbed K-means 9++w, as well as linear programming. The methodology integrates the CRISP-DM model and the K-means++ initialization strategy to overcome computational inefficiency and suboptimal initialization of traditional algorithms. This study analyzed data from various sources, such as railway track length, passenger and freight transport, and per capita GDP. The results demonstrate the effectiveness of Chebyshov and weighted Euclidean distance metrics in data processing and variable evaluation. Crucially, the proposed weighted Euclidean distance showed a superior performance, achieving the highest silhouette coefficient (0.3807) and the lowest Davies-Bouldin index (0.8174) in comparison with traditional distance variants. ABM optimization through automatic rules yielded time savings, while the normalization of scattered data improved coherence and computational performance. The implications of this study indicate the global potential of the railway construction campaign and a significant potential for the railway sector in Colombia. Our analysis of the findings, carried out with a global dataset, validates the model's applicability to the expansion of a B2B company beyond its local market. The limitations include the availability of public data and time and budget constraints.