EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
925. Investigating fairness in decisions involving ordinal classification - A COVID-19 case study
Invited abstract in session TA-6: Advancements of OR-analytics in statistics, machine learning and data science 12, stream Advancements of OR-analytics in statistics, machine learning and data science.
Tuesday, 8:30-10:00Room: 1013 (building: 202)
Authors (first author is the speaker)
| 1. | Edward Abel
|
| SDU | |
| 2. | Sajid Siraj
|
| Leeds University Business School, University of Leeds |
Abstract
Ensuring fairness is important when evaluating decision support systems aiming for impartial decision outcomes and trust within those being affected. During the COVID-19 pandemic, numerous nations implemented tiered restrictions as a localized strategy to balance economic activities against restrictions of movement independently in different geographical areas. However, such approaches sparked debate on the fairness of these restriction decisions. We examine fairness concerning the UK government's allocation system for tiered restrictions which can be modelled as an ordinal classification problem. For this, we collected and integrated data from multiple official sources to investigate potential inconsistencies, such as comparing the North and the South of England. We explore if there is inconsistency, by first training an ordinal classification model using only data from one geographical area, then testing the model with data pertaining to another geographical area. This helps us identify and measure classification errors in terms of underestimates and/or overestimates of predicted values compared to the actual values. Such analysis can be useful for exploring ordinal classification problems, both for post-hoc analysis of past decisions, and for providing transparency as part of future decision dissemination. Our approach could be applied to other ordinal classification problems, to explore fairness within other domains.
Keywords
- Analytics and Data Science
- Decision Support Systems
- Decision Analysis
Status: accepted
Back to the list of papers