{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/on-the-need-for-a-language-describing","title":"On the Need for a Language Describing Distribution Shifts: Illustrations on Tabular Datasets","arxiv_id":null,"date":"2023-09-26","proceeding":"NeurIPS 2023 11","authors":[],"abstract":"Different distribution shifts require different algorithmic and operational\n  interventions. Methodological research must be grounded by the specific\n  shifts they address.  Although nascent benchmarks provide a promising\n  empirical foundation, they \\emph{implicitly} focus on covariate\n  shifts, and the validity of empirical findings depends on the type of shift, \n  e.g., previous observations on algorithmic performance can fail to be valid when\n  the $Y|X$ distribution changes.  We conduct a thorough investigation of\n  natural shifts in 5 tabular datasets over 86,000 model configurations, and\n  find that $Y|X$-shifts are most prevalent.  To encourage researchers to\n  develop a refined language for distribution shifts, we build\n ``WhyShift``, an empirical testbed of curated real-world shifts where\n  we characterize the type of shift we benchmark performance over.  Since\n  $Y|X$-shifts are prevalent in tabular settings, we \\emph{identify covariate\n  regions} that suffer the biggest $Y|X$-shifts and discuss implications for\n  algorithmic and data-based interventions.  Our testbed highlights the\n  importance of future research that builds an understanding of why\n  distributions differ.","url_abs":"https://openreview.net/forum?id=PF0lxayYST","url_pdf":"https://openreview.net/pdf?id=PF0lxayYST","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"on-the-need-for-a-language-describing","repo_url":"https://github.com/namkoong-lab/whyshift","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[],"methods":[{"method_slug":"focus","method_name":"Focus"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}