Elements of a successful ML project

Also see data science questions to ask in an enterprise context.

  1. Are you solving a problem that stakeholders are actually interested in?
  2. Set reasonable goals and manage expectations in your business environment in order to not over-promise and under-deliver
  3. Have enough data of sufficient quality, because models are data hungry
  4. Prepare for maintenance and manage expectations! Models performance will decay over time.
  5. Develop with production in mind (MLOps)
  6. Have the right team, because ML is a cross-functional task and you need many different skills in your team.