S, R, and data science

Abstract

Data science is increasingly important and challenging. It requires computational tools and programming environments that handle big data and difficult computations, while supporting creative, high-quality analysis. The R language and related software play a major role in computing for data science. R is featured in most programs for training in the field. R packages provide tools for a wide range of purposes and users. The description of a new technique, particularly from research in statistics, is frequently accompanied by an R package, greatly increasing the usefulness of the description.The history of R makes clear its connection to data science. R was consciously designed to replicate in open-source software the contents of the S software. S in turn was written by data analysis researchers at Bell Labs as part of the computing environment for research in data analysis and collaborations to apply that research, rather than as a separate project to create a programming language. The features of S and the design decisions made for it need to be understood in this broader context of supporting effective data analysis (which would now be called data science). These characteristics were all transferred to R and remain central to its effectiveness. Thus, R can be viewed as based historically on a domain-specific language for the domain of data science.