⚖️

Fairness in Machine Learning

Final Project — Interactive Guide

Before you begin:
  • Your name and case study are linked. Once you begin, your case study is fixed for that name. To explore a different case study, return here and enter a different name.
  • Your responses are saved to your teacher as you commit them. They are visible to your teacher, not to other students.
  • Use the same name every session to pick up exactly where you left off.

Returning student? Enter the same name you used before — your previous responses will load automatically.

🎓
Case Study A: Education
Prioritising Students for After-School Support
A school district uses a machine learning model to predict which students are at risk of not graduating. Students flagged as high risk are offered after-school support programs. Resources are limited — not every at-risk student can be served.
Protected attributes: Race, socioeconomic status, English language learner status, disability status
🌊
Case Study B: Disaster Response
Prioritising Damage Assessment After a Natural Disaster
An emergency management agency uses a machine learning model to classify social media posts after a natural disaster — identifying which reports describe serious damage and should be prioritised for emergency response.
Protected attributes: Race, national origin, socioeconomic status, language — internet access and social media presence are proxy variables that correlate with these protected attributes, not protected attributes themselves
🏦
Case Study C: Lending
Bank Loan Denial Decisions
A bank uses a machine learning model to make loan denial decisions. Applicants above a risk threshold are denied credit. This connects directly to the German Credit and Adult Census work from the semester.
Protected attributes: Sex, race, age, marital status
✏️
My Own Scenario
Student-defined deployment context
You will define your own ML deployment context.
Protected attributes: To be defined by you