Follow good data management practices. These are domain neutral, but might have additional requirements in your domain. Researchers need to think about data management outside their domain and make their data FAIR for everyone in the project to use it.

[rndsrc]

SOP:

• Familiarize yourself with good data practices (list references), such as:
• Plan for your data to survive the duration of your project and beyond (remember: the person who is most likely to reuse “your” data is “you”!) (data storage: https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1005097; care of data: https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1003542)
  • Preparing data for a ML analysis is different from preparing data for publication. What needs to be included in datasets for ML?
  • Dealing with data streams
• Make a data management plan (which might be already prescribed by the funding agency); e.g. (https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1004525 “Ten Simple Rules for Creating a Good Data Management Plan”)
  • Data privacy and ethics
  • Be prepared to publish data from the moment you start collecting it
  • Make all changes to your data reproducible
  • Revise DMP as needed
  • Budget time (10%?) to spend documenting and following the DMP
  • List kinds of data generated and standards for collecting it
    • See NSF for key things to think about: https://nsf.gov/eng/general/ENG_DMP_Policy.pdf
• Learn about standards in your field (e.g., biosharing)
• Learn how to use metadata and identifiers and why
• Open formats (https://www.nature.com/articles/sdata201618 FAIR principles for data management)
• Special challenges with data types like images or sound