Software

I aim to develop my R packages following best practices in reproducible and testable software. These practices ensure the reliability and maintainability of my code, and support open science and reproducibility.

I use Git for version control, testthat for unit testing, Codecov to track test coverage, and GitHub Actions for continuous integration (CI) — including automated testing and CRAN checks during development.

R Packages

• ivgets package (CRAN (release) and GitHub (development)) for general-to-specific model selection and indicator saturation methods in two-stage least squares models. You can find an illustration below or visit the external app.
• robust2sls package (CRAN (release) and GitHub (development)) for outlier detection and testing in two-stage least squares models. Allows for various algorithms to detect outliers in the sample, conduct statistical tests for the presence of outliers, provides valid inference under the null hypothesis that there are no outliers, and tests for whether the effect of outliers in the parameter estimates are significant.

Illustration

This is a simple illustration of the outlier detection methods that I have been working on.