SimBaRepro

Simulation-based, finite-sample inference for privatized data

In 2024-25, I worked with Prof. Jordan Awan and Zhanyu Wang to turn their research on simulation-based reproducibility inference into an R package published on CRAN.

Illustration of the simulation-based reproducibility inference workflow
A high-level sketch of the inference workflow behind the package.

The package uses simulation-based methods to obtain p-values for hypothesis tests, confidence intervals, and multivariate confidence regions from privatized data. The multivariate regions can also be projected down to two-dimensional confidence regions for visualization.

Indicator array representation of a two-dimensional confidence region
Indicator-array representation of a two-dimensional confidence region.
Visualization of the two-dimensional confidence region
The same region visualized with the package's plotting tools.

The package page on CRAN has installation details, documentation, and the current released source.