Over years I came up with a quite standardized template. Though it usually degraded to copying a number of files from previous projects. With this template I try to formalize this approach.

cookiecutter is a tool to template new project files and folder structures. It still seems quite active and has templates beyond a single language it seems a promising home for this task.

These templates also contain some sample code for my reference how to achieve typical tasks. The most code is written in R files and derivatives like .Rmd and .Rnw. But I also make use of make for building the project, and additionale shell utilities like latexmk and pandoc. At some future point I will provide a setup script - at the moment you will need to weed through dependencies manually.

Install cookiecutter command line: pip install cookiecutter

Or with homebrew cookiecutter command line: brew install cookiecutter

Generate a new Cookiecutter template layout: cookiecutter gh:bdcaf/cookiecutter-r-data-analysis. It creates a minimal project that can be run typing make. Explore the Readme.md in the generated directory for more.

If your .cookiecutterrc is configured a quicker start is: cookiecutter gh:bdcaf/cookiecutter-r-data-analysis project_name="This is Magic"

Note Many functions used require that the project is a valid R package. The slug must be a valid package name, therefore it must only consist of characters, numbers and ., also it must start with a character. I wrote a simple replacement for spaces to dots, but for the rest you are on your own!

Besides cookiecutters dependencies it uses make, a perl script for dependency discovery.

Obviously a working R installation is also required. To have the automatic loading of code in R the devtools package is required. For the creation of reports knitr is required.

The exemplary work flow has plenty of dependencies. The main ones are:

• R with knitr, pander, dplyr, ggplot packages
• pandoc to convert into Word file
• latex for pdf generation
• curl to download the example data

This project is licensed under the terms of the MIT License