With increasing public expectation and scrutiny, along with developments such as the Algorithm Charter, operational algorithm projects must now include clear governance and good process.
The governance of operational algorithms is a rapidly evolving area. We are leading this evolution and aiding the whole sector by making our governance methodology available to everyone via the Model Development Lifecycle.
See how we've done this.
Ministry of Social Development (MSD)
Model Development Lifecycle
In product development projects, it’s common to have additional advisory boards alongside traditional project governance frameworks. For data science products, including operational algorithms, these are less common. This is not because they’re not needed, but because it’s a new field and governance of data science products isn’t yet well understood.
There are a lot of frameworks and principles that have been put forward by agencies in the past that are applicable to data science, but these are often targeted to a specific area and are theoretical-based. We worked with MSD to create the Model Development Lifecycle (MDL), a governance framework for operational algorithms. The MDL helps decision makers feel assured that the technical, legal, ethical, Te Ao Māori opportunities and risks have been well managed during the development of an operational algorithm.