liveBook · Manning

Mitigation

As explained in Delivering Machine Learning Projects: From design to deployment

A mitigation is an action or measure intended to reduce the likelihood or impact of a risk.

A mitigation is an action or measure intended to reduce a project's risk, likelihood, or impact. Effective planning identifies workable responses, estimates their effort, records uncertainty, and obtains accountable sign-off. section 3.2.3

Use mitigations as explicit project actions:

  • Identify proven mitigations and novel workarounds separately; price the work and attach a confidence level. section 3.2.3
  • Apply a larger risk premium to low-confidence estimates, and review high-impact risks that lack a credible resolution path. section 3.2.3
  • Record staffing, infrastructure gaps, scope changes, and proposed responses in the risk register; document decisions and share them with stakeholders. section 4.2 section 4.4.2
  • During emergencies, maintain backup options, assign actions, establish a reporting cadence, and continue prevention work after service is restored. section 10.2.4

Infrastructure or staffing gap

A promised production GPU server will not arrive for nine months. Record the infrastructure delta, then negotiate a non-GPU approach, change the provision, revise scope, or stop the project. If key staff are unavailable, escalate for replacements, replan delivery, or use contractors while reassessing the commercial case. section 4.4.2 section 4.2

Mitigation is not a promise that every risk can be removed. Some data, availability, or delivery constraints may remain unmanageable; the responsible response is to expose the uncertainty and let accountable stakeholders accept, change, or reject the proposal. section 3.2.3

How Delivering Machine Learning Projects: From design to deployment covers it

How deep it goes
Details identification, confidence pricing, escalation, documentation, scope changes, emergency reporting, and post-incident prevention.
Examples
Uses GPU delays, unavailable team members, replanning, contractors, emergency spreadsheets, backup plans, and reporting cadences.
What it assumes
Assumes a project with risks, a risk register, resource and infrastructure constraints, a budget, and accountable stakeholders.
What you can do after
Operate a risk register with priced responses, confidence levels, owners, escalation paths, documented decisions, and follow-up reviews.
How it connects
Connects risk registers with budgets, workarounds, scope changes, infrastructure, resourcing, monitoring, and stakeholder sign-off.

Also explained in

Compare how 4 books explain mitigation →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Delivering Machine Learning Projects: From design to deployment. The section links point to the parts of the book it draws on.