| tags: [ reproducibility ] categories: [meeting ]

reading club reproducibility

BES Reproducibility guidelines.

Good for YOU, and for reproducibility.

Uptake in education: Fiona new reproducibility subject at the University.

Setting up a project / programming

  • Archive old papers + code: local folder (Nick)
  • RANDOM SEED… forgetting to set. Compounded by long run-times and large file-sizes.

Accessibility… Barriers to startign work on reproducibility, but also, what are the bare minimum standards… -> Little changes each new time.

Statistical abstinence vs. safe statistics (avoid stats, and take to experts, OR tools to make safer). We are expected to wear lots of hats, good coding, good stats, field skills.

Version Control

Git - main purpose is NOT reproducibility. It’s main purpose is for collaborative development of operating system code. Not designed to be user-friendly for our purpose. Is genuineliy intimidating.

Combining git and dropbox. If you accidentally delete stuff without committing you’ve got another layer of dropbox.

COST of reproducibility - grand-writing, do you budget / account for time in implementing reproducibility. UK model: designated resources (people eg. software engineer). Known to everyone because is affiliated through an overseeing council.

Saves you time later!!! “technical debt”, programming term. Investing time now saves heartache later.

Other things not given time, but to be considered:

  • Reproducible reports
  • Debugging / defensive coding