P. Martínez Vaquero, E. López Cano, J. Martínez Moguerza
It is relatively easy to create statistical software using the R programming language. At the time this is written there are 18884 contributed packages in the CRAN repository. Usually, the focus when creating an R package is on the statistical method to be implemented. Very often engineering software good practices or software quality techniques are disregarded or ignored. It is time for statisticians who create statistical software to incorporate these techniques into their workflows in order to have high quality statistical software. One of this techniques is unit testing. Unit tests can be created for each function in the package. Tests are automatically run when building the package, and we can test the whole package when needed. This is important when creating the function for the first time, but also when maintaining the package to avoid breaking code that previously worked. Our experience implementing unit testing in the SixSigma R package will be shown.
Palabras clave: R, software engineering, statistical software, unit testing, six sigma, quality
Programado
GT20 Software y Computación para Estadística e Investigación Operativa (SOCEIO)
9 de junio de 2022 12:00
Salón de Grados
