Abstract

Agile organizations require the ability to form balanced, cross-functional teams rapidly, particularly during employee onboarding phases when information is limited and decisions must be made quickly. This study proposes a machine-learning-driven approach for workforce segmentation that supports agile squad formation using only onboarding-stage employee data. The objective is to identify interpretable capability-based personas that can inform early staffing decisions without relying on performance outcomes or post-hoc evaluations. A publicly available employee onboarding dataset was analyzed using unsupervised clustering techniques. Feature representation focused on workforce capability proxies and onboarding logistics, including educational attainment, qualification level, trade specialization, geographic information, and joining time. A unified preprocessing pipeline was applied to ensure consistent handling of numeric and categorical attributes, enabling fair comparison across clustering algorithms. Three clustering paradigms—KMeans, Gaussian Mixture Models, and Agglomerative Hierarchical Clustering—were evaluated across a range of cluster counts. Cluster validity was assessed using internal metrics, namely Silhouette Score and Davies–Bouldin Index, complemented by inertia-based elbow analysis for KMeans. The results consistently indicate that the data supports only a small number of meaningful segments, with optimal solutions emerging at low cluster counts. Visualization using two-dimensional principal component analysis further reveals overlapping but structured workforce patterns, suggesting gradual capability transitions rather than sharply separated groups. The findings are discussed in the context of agile workforce management, emphasizing the interpretation of clusters as soft personas rather than rigid categories. These personas can be combined flexibly to form balanced squads, supporting rapid decision-making while preserving adaptability. The proposed framework demonstrates how unsupervised learning can enhance transparency and speed in onboarding-related staffing decisions, offering a practical contribution to agile HR analytics and workforce planning research.

Keywords: Agile Workforce Management, Employee Onboarding, Clustering Analysis, Workforce Segmentation, Machine Learning

How to Cite:

Ayuningtyas, A. & Wulandari, R. N., (2026) “Capability-Based Workforce Clustering for Agile Cross-Functional Team Formation in Onboarding Programs”, Agile Management 1(4), 226-241. doi: https://doi.org/10.63913/am.v1i4.138