Hi all,
I have released stratifyR 2.0-1 on CRAN and wanted to share it here in case it is useful, and to get feedback from this community.
The package helps design stratified surveys when the stratification variable is continuous, like income, production, or farm size. You give it your data, the number of strata, and the sample size you can afford, and it returns the optimal stratum boundaries and the sample size for each, which can cut the variance of an estimate a long way compared with simple random sampling.
New in 2.0-1:
- Three solvers, a globally optimal Dynamic Programming method plus two faster derivative-free ones.
- Cost-constrained allocation for unequal per-stratum costs.
- A design-comparison function that reports the efficiency gain.
- 2D, 3D and interactive plots.
- A Shiny app, launched with stratifyRApp(), for the whole workflow without code.
Install with install.packages("stratifyR"). I wrote a short walkthrough here: RPubs - stratifyR 2.0: A Major Update for Optimal Survey Stratification
I would love to hear how others approach stratum boundary selection in practice, and any feedback on the package or the app.
Thanks.