Model Development Lifecycle
Government agencies in New Zealand use data to help inform, improve, and deliver services to people every day. The use of data must be carefully managed to ensure that it is safe and effective.
Model Development Lifecycle and Operational Algorithms
The Ministry of Social Development (MSD) created the Model Development Lifecycle (MDL) as a practical guide to help manage new and emerging uses of data in an operational setting (operational algorithms).
The MDL is designed to work with an organisation’s internal processes and frameworks. It supports users to manage and document their approach to issues around privacy and ethics.
Frameworks help manage the use of algorithms and provide guidance for:
- striking the right balance between privacy and transparency,
- preventing unintended bias, and
- reflecting the principles of the Treaty of Waitangi.
The MDL helps MSD meet our commitment to the following:
- Te Tiriti o Waitangi / the Treaty of Waitangi
- Algorithm Charter of Aotearoa New Zealand
- Data Protection and Use Policy
The MDL is supported by, and works with, the Ministry of Social Development’s Privacy, Human Rights, and Ethics framework.
The MDL is an open-source document, and is intended to be used by other agencies. The Ministry will update these guides as we learn more about the safe and effective use of algorithms. If you have any feedback, you can email us at research@msd.govt.nz
An example of using the MDL is the Youth Service for young people Not Employment, Education or Training (NEET), an automated system for referring young people to Youth Service.
Using the Model Development Lifecyle
The Model Development Lifecycle contains three documents. These guides provide decision makers with assurance that technical, legal, ethical and te ao Māori opportunities and risks are managed throughout the lifecycle of an algorithm.
User guide
This document is a guide to roles and responsibilities for the development of operational algorithms.
Governance guide
Provides examples of governance to support decision-making for operational algorithm products.
Data science guide for operations
Supports data scientists to apply their academic knowledge to solve real-world problems.