I have been using rmarkdown in my research and really enjoyed it. I also tried to get my head around shiny and ggvis. I can understand what they do but, compared to rmarkdown, dplyr, and ggplot2, they seemed less relevant. Yesterday I went to the R-Day of the Strata+Hadoop meeting and the presentations made by Winston Chang and Garrett Grolemund really changed my opinion. And the timing is perfect: now I have strong incentive to incorporate both technologies into my Advanced Analytics course in the Spring of 2015.
According to Hadley Wichham, the author of dplyr and a number of other excellent packages, the next version of dplyr will have a function for recoding.