Skip to content

Background

The purpose of these continuously updated notes is to outline points critical to the success of machine learning dependent projects. The expectation is that prospective clients/collaborators will thoughtfully, carefully, consider the time commitment, collaboration commitment, cost, range of expertise, project details, etc., that are required for a machine learning dependent project; and subsequently make well informed decisions about whether to proceed with a project.


A project's details are especially critical to conducting a well-informed technical feasibility & economic viability assessment before proceeding. It is also important to beware that the data science aspect of the project is only a part of the project, ignoring this leads to project failures due to, e.g.,

  • Under-planning.
  • Ambiguous project scope/design details.
  • Project pilots that are neither informed by nor tests the deployment goal.
  • Underestimating, or ignoring, cost.
  • Ignoring the critical importance of intensive collaboration between the business, data science, operations, etc., teams.


One of the first sections of these pages is Deployment Goal. Colleagues may question why. Here are a few references that capture our experience of why business data science projects fail, and the advantages of starting with ``the end in mind''