jtleek / capitalin21stcenturyinr Goto Github PK

Hi,

I'm in the process of creating beamer slides for Piketty's book.

https://github.com/ptoche/capitalIn21stCenturyinR

To begin with, I am going to replicate the powerpoint slides Piketty used at a Sao Paulo presentation in December 2014 (because they are the latest slides I found). I have just completed the short version. In the process, I have refactored some of the code. However, I would welcome suggestions about best practice so that my code may be subsequently merged into the main project.

1. The data is defined in the preamble of the ".Rnw" document, which is the master file for the 'Slides' project. I have named dataframes used to produce the book figures according to the figure number in the book, e.g. df15.1 for the dataframe used to produce figure 15.1. Sometimes the data comes from a worksheet labeled TS15.1 in the original excel files, but sometimes it comes from a different sheet, because the same sheet might be used to produce several figures and/or because something is unclear in the naming scheme used by Piketty in his worksheets. For this reason, when I first load the data I name the dataframe ts15.1 and later change it to df15.1 or df15.2, etc. as the case may be.
2. I have not used too many special functions to produce the plots because I think that to be most useful it should be reasonably easy for anyone to cut/paste the code for a single figure. There are 2 exceptions to this: a shapePalette and a colorPalette which I define in the preamble. I have also defined a black and white scheme to produce figures quite close to the original book style.
3. In this incarnation of my code-in-progress, I am producing the graphics on the fly as the beamer is produced. In a later refactoring, it may be desirable to produce all the graphics by Chapter, as the current repository does, and simply use includegraphics in the beamer presentation, pointing to the location of the figures in the book repository.
4. Once all the book figures have been produced in a consistent way and dataframes named in an intuitive way, e.g. df15.1, I would save the dataframes in a .rda file. And perhaps create a 'piketty' package with the stored data.

Any advice/suggestions?