So just a little bit about R Markdown, is R Markdown is built into R Studio.
And basically it allows you to create documents,
particularly HTML documents from within R.
But in such a way that it embeds the code and the results of the code and
the plots and everything else that you're doing from your analysis Into the document
in one big reproducible mish mash.
And that's incredibly useful, because the changes this old workflow
of having some analytic R code that is maybe documented, but probably not.
And then a separate document, say a Google document or a Word document, or whatever,
a LaTeX document or whatever.
And the way you're combining them is by copying and pasting or
saving files manually and inputting them into the documents.
Same thing with presentation, you might have a PowerPoint presentation and
you're running your code over here and then pasting or
copying those things into the presentation.
And that leads to issues with version control, it also leads to the problem that
the person you're giving the presentation to if you're handing them the file
can't necessarily reproduce all the figures and graphs in it.
So, again, this is just in general a better way to do things if
you're concerned with having things like reproducibility and
having a nice consistent structure for generating your presentations.
I can imagine an instance in say a corporate setting
where a particular report has to be given routinely.
Say every couple of weeks or something like that, from data,
important data that's being generated for the company.
And instead of redoing that presentation and having the error prone process of
copying and pasting each time, the idea of having a script that would generate
the presentation with the latest numbers in a way that everything was traceable and
version controlled would be an incredibly useful thing.
So one way to think about R Markdown in creating presentations like this is as
a useful way for creating recurring presentations or automated presentations.
But, it has a lot of uses and I want to emphasis this idea of reproducibility.
So this has received a lot attention here,
we talk about it a little bit in the slides.
This has received a lot of attention in the academic community now,
because mostly, with regard to papers, people have put out papers and
then independent groups have just tried to reproduce the results of the papers.
Not redo the experiments, just take the same data sets and
try to reproduce the analysis and it turns out that it's often not possible and
it creates disputes over who's right the first people or the second people and
now a third group did it and they can't reproduce it, but
they get something different than both the other two people.
So having a document that has the code
embedded in it really improves the status of reproducibility of the document.
It's not perfect, because there's probably some external files or
datasets that you have to call, some libraries.
There's probably differences between computing architectures and
other things that may impact your results.
But by and large this goes leaps and
bounds towards creating more reproducible results in your documents.
Okay, so hopefully you've heard a lot of this before when you were talking about
reproducability as it pertained to Nitter and things like that,
in your reproducible research class that's part of the data science specialization.
But if you're taking this class in Ovo, then we're just going to walk you through
a couple of example, focusing mainly on presentations.
An in this class, we focus on presentations just because
we like to think of this class as a products class, so
the idea is that you're generating product.
And that your presentation will be your pitch for the product.
And we want your pitch to be created in one of these reproducible formats.
Okay, so next we're just going to start going through some examples.
Brian Caffo, PhDProfessor, Biostatistics
Jeff Leek, PhDAssociate Professor, Biostatistics
Roger D. Peng, PhDAssociate Professor, Biostatistics
