[R] R report generator (for Word)?

[Paul]

On 01/01/12 15:50, Michael wrote:
> Happy New Year all!
>
> I am looking for a good solution for keeping record of my experiments -
> could you please help me?
>
> My work is about analysing data... My current work-flow:
>
> 1. Everyday my bosses give me some small steps/tasks for analysing data -
> which are parts of one bigger/whole project.
> 2. Everyday I send tens of emails to bosses/colleagues to report my
> findings in each step.
> 3. Bosses/colleagues often respond to my findings in real-time and suggest
> new experiments/steps and ask "what-if" questions.
> 4. I often have to manually copy and paste the results from R console and
> put them into an Excel and decorate a bit and send out.
> 5. Every one week and 2 weeks, we need to present to more senior bosses
> with more nice-looking presentations which is a summary of our findings in
> those 1-2 weeks. It's this time that is most chaotic because my colleagues
> and I have to dig into all the hundreds of emails in the past 1-2 weeks and
> copy and paste and organize those data again and make a nice overall
> summary for presentation...
> 6. As I am a hard-working guy, I myself often run my own random/ad-hoc
> experiments using out-of-work time and whenever I have interesting
> findings, I will send to immediate bosses and colleagues to seek their
> comments.
> 7. All these experiments are in fact variations of different versions/ideas
> of one big/whole project. Lets say in one big project bosses/colleagues and
> I have come up with a few big ideas, then we have a few sub-projects:
>
> MyProjectIdea1
> MyProjectIdea2
> ...
> MyProjectIdeaN
>
> And each idea has a few variations, mostly are for answering "what-if"
> questions by varying the parameters here and there ...
> For example:
>
> MyProjectIdea1_Variation1_WhatIfParam1ChangedTo1.2?
> ...
> ...
> etc.
>
> 8. Most experiments run tens of minutes to many hours... and some of them
> have to run on Linux, and some others can be run on Windows. Fortunately we
> have universal paths accessible on both Windows and Linux, so those won't
> be problem...
>
> 9. Because of the time-consuming nature of these experiments, I also save
> the images as "rData" whenever I can. However, it's necessary to keep track
> of the context where these data were generated. Otherwise even the records
> of these images won't help recall the scenario we have run...
>
> ---------------------------------------------------
>
> Keeping track of these changes and all kinds of "what-if"s now becomes
> increasingly a problem for me.
>
> Some times in order to respond to a query, although I have done it before
> already, but because I didn't keep record and save the result, or even
> though I have saved the memory image yet I am not completely sure about the
> "cleanness" of the results/data,I have to redo it and wait for another few
> hours.
>
> Is there a way that I can manage these whole processes better and be more
> productive?
>
> I have been digging and thinking about this for while and I guess Sweave is
> the right way to go?
I would agree that Sweave is a good way to organise these kind of 
analysis that you do repetitively. I would also recomend you look at 
some of the caching methods available (CacheSweave 
http://cran.r-project.org/web/packages/cacheSweave/index.html ) will 
allow you to skip chunks of the analysis that haven't changed since the 
last run.
> The problem for Sweave is that it's hard to make Latex generated pdf
> appealing to business managers... so if I keep records in Sweave/Latex for
> my own record/benefit (that's already a big benefit)... I still need to
> somehow manually copy/paste the data from Sweave/Latex/pdf into
> Word/Excel/Powerpoint in order to make a nice presentation...
>
> I know there are some Open Office and Word version of Sweave... the problem
> is that I couldn't find many demonstrations on these topics and my question
> is: are they good and can they fulfill what we needed?
>
For Open Office have a look at the odfWeave package ( 
http://cran.r-project.org/web/packages/odfWeave/index.html ) that 
functions in a similar way to Sweave but starts with and produces an odt 
Open Doc text file.  I have used it in the past but not recently, 
although I vaguely remember seeing a post recently that the current 
version doesn't compile on Windows, so that may curtail it's effectiveness.

regards,

Paul.