Skip to main content

Skills application

Instruction and application
Complete

Designing an ML data strategy​

In this skills application, you will put into practice the concepts from this unit by analysing a real-world scenario and designing a data strategy that is compliant, transparent, and fair.

Success criteria​

To successfully complete this skills application, you must:

  • Identify relevant regulations and ethical risks associated with student data.
  • Apply an appropriate governance framework (e.g., AREA or SAFE-D).
  • Propose an actionable data strategy (minimisation, tracking, retention).
  • Recommend practical data quality and fairness checks.

Context​

You are part of a cross-functional data team at a public sector agency developing a machine learning model to predictstudent performance outcomes. The model is trained on a combination of behavioural, demographic, and academic data sourced from multiple school districts.

Some of the data is sensitive, including ethnicity, socio-economic status, and disciplinary records. Education advocates have raised concerns regarding data retention, fairness, and transparency. Your task is to evaluate these concerns and design a compliant and ethical data strategy.

Instructions​

Follow the prompts in the form below to complete your analysis. Completing this activity will “unlock” the solution example on the following page.

Regulatory and ethical analysis • Identify one data privacy regulation relevant to this student data context. • Briefly explain how it impacts how the data should be collected, stored, or used in the ML pipeline.
0 words
Governance framework application • Select either the AREA or SAFE-D framework. • Describe how two of its principles would guide responsible data use in this project (e.g., explainability, fairness, accountability).
0 words
Compliant data strategy • Propose three actions that support data minimisation, metadata tracking, and retention limits. • Describe how access to the data would be managed and how audit-readiness would be maintained.
0 words
Data quality and fairness plan • Recommend two quality control checks and one fairness check to embed in the pipeline. • Briefly explain how you would document bias detection and respond to identified issues.
0 words
In progress