One area where clearly GUI methods are preferable to command line methods in R, is data input. There is no need of learning read.csv or read.table when these options are only two clicks away in any R GUI. For academics/students there is a definite need to easily access
datasets from attached packages just as it is a need for business analysts to access databases with a few clicks than learn or read pages of pdf on RODBC. However some GUI (like Rattle) need data only in data frames, rather than list or arrays-this limits R’s flexibility. These are my views but you can see and compare views of data input in R Commander, Rattle and Deducer.