David, Lee Wilkinson has already done this (remarkably well). His package was called "A Second Opinion" (ASO) but I think it has been taken over by a larger company and has thus a price tag. But check with Lee. Basically you inout data and tell it what you want to know and it does the appropriate analyses and outputs the result as a written report complete with tables, graphs and references. By Lee's count it does a better job than 90% of all statisticians. The program has become a favorite to many Chinese statisticians who find that it is writing the intro and results sections of their papers for them.

Hope this helps.

Howad

Thanks a bunch Howard. I wrote Lee when I started doing this and he never wrote back. I will recontact him about this.

This idea has been implemented in open source as well. It is called "literate programming". The idea is "weaving" a document, i.e. writing prose with embedded source code intersections, then evaluate the code and end up with prose featuring the desired outcomes of these expressions.

A practical example would involve embedding R code in the context of an analysis report, evaluating it, and typeset the result with LaTeX into an APA formatted pdf or html file with the actual numbers in-line. Plots and tables are produced similarly. The user is in complete control of whether the source code, warnings, errors, and even results, show up in the document. For more demanding documents, such as a thesis, one can also only produce a tex document, in order to apply finer adjustments, such as multipage tables . The tex file can then be compiled into pdf, as well as other formats.

https://support.rstudio.com/hc/en-us/articles/200552056-Using-Sweave-and-knitr