Quantitative Approaches in Longitudinal Skill and Competence Development

The ongoing evolution of work, driven by technological advancement, demographic changes, and societal values, necessitates deeper insights into career trajectories, skill demands, and job matching. This research addresses these dynamics through three original studies. First, Dynamic Time Warping was employed to cluster soccer players’ career paths across the European Big-5 leagues, revealing significant predictors in early career phases, such as years in youth academies and debut leagues. Second, skill demand forecasting was advanced by incorporating synthetic data generated through TimeGAN, which enhanced the prediction accuracy of Long Short-Term Memory (LSTM) models and demonstrated the robustness of synthetic data in handling unseen data within dynamic industries. The third study explored the use of an ensemble deep neural network model combining TextCNN and LSTM to improve the accuracy of job predictions. The ensemble model outperformed other models with an F1-Score of 89% on an 89-class classification task, demonstrating its efficacy in leveraging
high-level feature representations. Collectively, these studies contribute to the fields of sports analytics, workforce planning, and human resources by providing innovative methodological approaches and demonstrating practical applications. The findings offer actionable insights for identifying promising talent, forecasting emerging skill needs, and optimizing job placement strategies, while also presenting avenues for future research across various domains.