Research Reproducibility (RR)

[Page under construction]

Goodman et al. (2016):

• Methods
• Results
• Inferential

Five creative ways to promote reproducible science

Version control

• Git

Reproducible environments

• New!! Best practices to RR from the perspectives of a Kaggler
• R and Python with pven, renv, venv: see blog from Nov 2023
• Python:

  pip freeze > requirements.txt
  

• Dockerfile
• YAML is a human readable language
  • Often used to write configuration files

    ---
    title: "Reports on alerts"
    author: "Winnie Mo"
    date: "2024-APR-05"
    format: html
    jupyter: python3
    ---  
    

Publishing platforms

• Dryad
  • A community-owned resource
  • Python package dryad2dataverse
• Open Science framework.io

Coding platforms

Free with limits

• Kaggle
• Digital Ocean
• Google Colab
• Jupyter.org
• ...

Sponsored accounts

Option 1: Servers

Replace yourusername with your's

• https://sfu.syzygy.ca/jupyter/user/yourusername/lab
• https://ubc.syzygy.ca/jupyter/user/yourusername/lab
• https://jupyterhub.dataspace.copernicus.eu/

Option 2: Digital Research Alliance

Digital Research Alliance; new users require sponsorship

Option 3: UBC Advanced Research Computing

Digital Research Alliance; new users require sponsorship

Recommended practices

• markdown

Example workflows

R

# Makefile (target: required_files)

manuscript.pdf: manuscript.Rmd simulated_data.csv 

Rscript -e 'rmarkdown::render("manuscript.Rmd")' 

simulated_data.csv: simulate.R

Rscript -e 'source("simulate.R")' 

Source: MDPI 2020

References

• Davos: A Python package ‘‘smuggler’’ for constructing lightweight reproducible notebooks, Paxton C. Fitzpatrick, Jeremy R. Manning 2024 article
• Reproducible data science - Python

Return to Home