Conference Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 125 - Practical Recommendations for Prediction Modeling That Advance Innovation
Type: Invited
Date/Time: Monday, August 8, 2022 : 10:30 AM to 12:20 PM
Sponsor: Section on Statistical Learning and Data Science
Abstract #319258
Title: Disseminating Prediction Methods: Avoiding Computational Bottlenecks and Developing User-Friendly APIs
Author(s): Byron Casey Jaeger*
Companies: Wake Forest School of Medicine
Keywords: R package; C++; Rcpp; computation
Abstract:

The availability and 'user-friendliness' of software disseminating novel methodology are essential for getting it into the hands of interested investigators. Statisticians have traditionally written R packages to disseminate novel methods to broader audiences. However, R programs may run inefficiently on large datasets or when implementing methods that use high amounts of recursion. Additionally, as few statisticians have formal training in software design, these R packages may be difficult for new users or unreliable in certain settings. This presentation will cover best practices for R package development using Rcpp, which allows R users to implement their methods in C++. Examples will show how to set up a new package using Rcpp, how to maintain unit tests that ensure the package does not fail in unexpected ways, how to design user-friendly functions, and how to use libraries in C++ to obtain 10 to 100 times faster code versus traditional R scripts.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2022 program