It’s an iterative process

When you open a bad folder

Why

If you have to do something once, you are going to have to do it 1000 times

• analysis is iterative
  
• requests/collaborators are unpredictable
  
• data grows

If you can’t reproduce what you did, it’s impossible to find the mistakes
If you can’t reproduce what you did, it is not credible

Why

• If you can reproduce what you did, others can too
  
• They can either use the information to perform similar analyses, or build upon all the work you did

The best thing you can do is to start every project with the goal and mindset of reproducibility.

Stop working in the dark. Realize that everything you do will eventually be seen.

Steps to reproducible research

Everything with a script

• A script is a sequence of code that performs a specific job
• Pointing, clicking, dragging, dropping, copying, pasting are not reproducible
• Code tracks and records every step you take
  
• Get data -> convert file -> edit file -> clean data -> analyze data
• Still possible to write terrible scripts though

Organize your data and code

• Use one directory (local) and one repo (GitHub) for one project
• Put code and data in separate subdirectories
  

Organize your data and code

Organize your data and code

• Use relative paths that refer only to the project directory and not to your specific computer
  Good: images/friends.gif
  Bad: C:/Users/rdietzel/Documents/materials/week_02/images/friends.gif
  This will save your friends a lot of trouble when they run your code on their computer

Organize your data and code

Write ReadMe files (component of metadata)

• The ReadMe file is the first to be seen or sought
  
• Describe the files in your folder, especially which order to run them in
  
• Leave a map for those that come after you, or your future self
  

Automate the process

• Aim to write a script of your scripts
  
• This also helps you mentally organize the steps you need to take in your project
  
• Rerunning your project happens with the touch of a button
  
• Do months of work over and over and over whenever you need to
  
• Broman recommends GNU Make, but R is sufficient for this course

Turn scripts into reproducible reports

• Include enough text in your scripts that future you and anyone else will know why you did what you did
• We will devote time to reproducible reports later in the semester

Turn repeated code into functions - more later

• A function is a command made up of other code, e.g. mean, standard deviation
• Your code will be a lot easier to read and faster to write
  
• Package up these functions to share and use later

Use version control

• You are already doing this with GitHub!

Totally worth spending class time on file naming

Naming things

File names should

be machine readable
be human readable
play well with default ordering

Machine readable means

Regular expression and globbing friendly

• Avoid spaces, punctuation, accented characters, case sensitivity

Easy to compute on

• Deliberate use of delimiters

Globbing

Human Readable

Plays wells with default ordering

Not cool to manipulate default ordering with ill intentions

Plan your naming scheme

• Picture what your files, folders, project, and degree program will look like
  
• Name your file when you create it, not just before you need to run out of your office