Small area estimation improves on direct survey estimates by sharing information across areas with limited samples. The talk presents a framework for combining these methods with machine learning. It introduces learners for different amounts of available data, describes how to choose among them, and compares their performance with classical alternatives.
Session “Survey sampling: New methods and applications”, Room 20 (CO126).
The conference takes place 12–14 December 2026 at the Wilhelminenhof campus of HTW Berlin. The talk is listed as A1867 in the programme.