The Rfuncs Project
Why Is the Rfunc Project Not a CRAN Package?
Experienced R users may be troubled to find that Rfuncs are not formatted and distributed like those in the nearly 8000 CRAN Packages that were listed as of February 2016. Why have I not (yet) conformed?
The short answer
CRAN packages are no-doubt easy for users to install, but when I peek inside them, their programming style often falls way short of what Kernighan and PJ Plauger instilled in me, and their documentation and examples make for difficult learning. The typical Rfunc file is composed of (1) raw source code, (2) ample documentation, and (3) in-depth, realistic examples, all painstakingly crafted. Because the Rfunc structure and style worked so well for my students, I see no compelling reason to change.
That said, R experts, in particular my long-time friend (and former University of Tennessee colleague) Bob Muenchen, have told me that I can develop My Way and still conform to how CRAN Packages work. I hope to do that some day, but for now that's just something else that will hinder developing all the Rfuncs on my to-do list, which seems to expand and not contract.
more ...
The Rfunc Project grew out from my quest at Case Western Reserve University (CWRU) to deliver the most modern courses in statistical science that I could. In teaching, learning, and practicing statistics, the motor and drivetrain is solid software coupled with lucid documentation and rich examples.
Most of my students were pursing graduate degrees in subject-matter fields in biomedicine. Few were savvy at programming, and the vast majority had never used R. Inspired by a lecture given by W. Edwards Deming, in 1985, I then stopped assigning mundane homework exercises or giving examinations. Instead, for nearly 30 years, my students created and carried out real/realistic projects In their area of interest and presented them in small seminar sessions held outside of regular class times, with me openly questioning and criticizing in a “tough love” manner. But such pedagogy only works in statistics courses when everything is tied directly to software that is easy to learn and use. For various reasons, I found myself writing my own R functions. Almost all student projects were good and many were outstanding—better than my lectures! This format induced learning that sticks, forced me to stay modern and creative, and made teaching a source of great pleasure and satisfaction. Oh, how I miss it.
In retiring from CWRU in 2014 and moving to coastal North Carolina, I decided that my next professional quest would be to refine these functions and distribute them free and open-source. In the mid-1990s, I made a similar decision and freely distributed a huge SAS module for sample-size analysis called UnifyPow. It proved to be a hit and led to SAS Institute producing PROCs POWER and GLMPOWER, which I consulted on formally, intensively, and joyfully with Dr. John Castelloe. If The Rfuncs Project achieves even 5% of that success, it will be worth it.
The Rfunc Project grew out from my quest at Case Western Reserve University (CWRU) to deliver the most modern courses in statistical science that I could. In teaching, learning, and practicing statistics, the motor and drivetrain is solid software coupled with lucid documentation and rich examples.
Most of my students were pursing graduate degrees in subject-matter fields in biomedicine. Few were savvy at programming, and the vast majority had never used R. Inspired by a lecture given by W. Edwards Deming, in 1985, I then stopped assigning mundane homework exercises or giving examinations. Instead, for nearly 30 years, my students created and carried out real/realistic projects In their area of interest and presented them in small seminar sessions held outside of regular class times, with me openly questioning and criticizing in a “tough love” manner. But such pedagogy only works in statistics courses when everything is tied directly to software that is easy to learn and use. For various reasons, I found myself writing my own R functions. Almost all student projects were good and many were outstanding—better than my lectures! This format induced learning that sticks, forced me to stay modern and creative, and made teaching a source of great pleasure and satisfaction. Oh, how I miss it.
In retiring from CWRU in 2014 and moving to coastal North Carolina, I decided that my next professional quest would be to refine these functions and distribute them free and open-source. In the mid-1990s, I made a similar decision and freely distributed a huge SAS module for sample-size analysis called UnifyPow. It proved to be a hit and led to SAS Institute producing PROCs POWER and GLMPOWER, which I consulted on formally, intensively, and joyfully with Dr. John Castelloe. If The Rfuncs Project achieves even 5% of that success, it will be worth it.