WFC
Provides a disciplined precheck, execution, and diagnostics workflow for survey weighting and raking. Weight construction requires design-only data, a verified external target, outcome-blind planning, and human approval before weight locking, with bilingual reports for decision and statistical audiences. Converts calibrated and replicate weights into standard survey designs, provides optional broom-style result projections, and records serializable production pipeline provenance. Supports fixed, predeclared soft calibration tolerances and categorical entropy balancing from verified margins, plus panel attrition weighting, high-influence unit diagnostics, Fay's balanced repeated replication, and opt-in parallel execution for long runs. Calibration methods follow Deville and Saerndal (1992) <doi:10.1080/01621459.1992.10475217>, and entropy balancing follows Hainmueller (2012) <doi:10.1093/pan/mpr025>.
This page was built for software: WFC
