[2510.25974] Who Leads? Comparing Human-Centric and Model-Centric Strategies for Defining ML Target Variables

Computer Science > Human-Computer Interaction arXiv:2510.25974 (cs) [Submitted on 29 Oct 2025 (v1), last revised 29 Mar 2026 (this version, v2)] Title:Who Leads? Comparing Human-Centric and Model-Centric Strategies for Defining ML Target Variables Authors:Mengtian Guo, David Gotz, Yue Wang View a PDF of the paper titled Who Leads? Comparing Human-Centric and Model-Centric Strategies for Defining ML Target Variables, by Mengtian Guo and 2 other authors View PDF HTML (experimental) Abstract:Predictive modeling has the potential to enhance human decision-making. However, many predictive models fail in practice due to problematic problem formulation in cases where the prediction target is an abstract concept or construct and practitioners need to define an appropriate target variable as a proxy to operationalize the construct of interest. The choice of an appropriate proxy target variable is rarely self-evident in practice, requiring both domain knowledge and iterative data modeling. This process is inherently collaborative, involving both domain experts and data scientists. In this work, we explore how human-machine teaming can support this process by accelerating iterations while preserving human judgment. We study the impact of two human-machine teaming strategies on proxy construction: 1) relevance-first: humans leading the process by selecting relevant proxies, and 2) performance-first: machines leading the process by recommending proxies based on predictive performance. ...