LAS 6292: Data Collection & Management

Reproducibility

           

Why should we practice

‘Reproducible Research’?

Why ‘Reproducible Research’?

• Have you ever tried to reproduce someone else’s data analysis?

• Have you ever tried to repeat your own (old) work?

• What made it easy/hard to do so?

• What would have to happen if you had to extend the analysis further?

• If you caught a data error how easy/hard would it be to re-create the analysis?

• What would happen if your collaborator is no longer available to walk you through their analysis?

• What if you couldn’t remember the steps?

A Data Sharing Snafu in 3 Acts

Why ‘Reproducible Research’?

Scientific Integrity:

• Rigor, trustworthiness, and transparency
• Allows others to verify analyses, find & correct mistakes

Efficient Research:

• Helps remember HOW and WHY you did something.

• Speed up this project: Automation makes it faster (and easier) to redo analyses. Time saved can be spent doing other stuff.

• Speed up future projects: Can use scripts from one project for another one.

Community Good:

• Others can learn from your approach & methods

The Reproducible Research Workflow

Reproducible Data Processing: Let’s Practice

Reproducibility = Recipies