Changelog

Source: NEWS.md

samplyr (development version)

Core Grammar

• Five verbs + one modifier for composable survey sampling designs: sampling_design(), stratify_by(), cluster_by(), draw(), execute(), and stage().
• Frame-independent design pattern separating specification from execution.

Sampling Methods

• Equal probability: SRS (with/without replacement), systematic, Bernoulli.
• PPS methods: systematic, Brewer, maximum entropy, Poisson, multinomial, Chromy (PMR).

Allocation Methods

• Proportional, equal, Neyman, and optimal allocation.
• Custom allocation via data frames.
• Minimum and maximum sample size constraints (min_n, max_n).

Features

• Multi-stage and multi-phase sampling with automatic weight compounding.
• Certainty selection for PPS methods (absolute and proportional thresholds).
• Control sorting with serp() for hierarchic serpentine ordering.
• Partial execution for operational sampling workflows.
• Synthetic survey datasets from African countries for learning and testing.

Survey Integration

• as_survey_design() converts tbl_sample objects to survey::svydesign() for variance estimation and analysis.
• joint_inclusion_prob() computes exact joint inclusion probabilities (π_kl) for PPS WOR stages via sondage::up_*_jip().
• summary.tbl_sample() provides per-stage stratum allocation tables and weight diagnostics.