Speaker
Description
"People Analytics has emerged as a pragmatic application of data science to socio-technical systems in which human behavior, incentives, and organizational structures interact in complex ways. This talk presents two applied modeling problems that illustrate both the potential and the limitations of quantitative approaches when the subjects of analysis are people rather than physical or purely transactional systems.
The first case explores the probabilistic estimation of voluntary attrition risk among high-potential employees over a six-month horizon. Rather than treating attrition as a deterministic outcome, the problem is framed as probability estimation intended to support decision-making under uncertainty. The discussion focuses on problem formulation, data considerations, model choice, and the practical interpretation of predicted probabilities in organizational contexts.
The second case examines the use of multivariate linear regression models for pay equity analysis. Although methodologically simple, this application raises important questions around model specification, interpretation of coefficients, and the communication of results to non-technical stakeholders. The talk emphasizes how statistical outputs are translated into actionable insights while operating under legal, ethical, and organizational constraints.
Taken together, these two use cases position People Analytics as a domain where standard statistical and machine learning tools are embedded in complex adaptive systems. The models are not presented as definitive answers, but as decision-support instruments whose value depends on context, governance, and an explicit acknowledgment of uncertainty and human agency."