Abstraction and Prediction Algorithms: A Harm-Reduction Framework

Simple item page
aut.embargoNoen_NZ
aut.thirdpc.containsNoen_NZ
dc.contributor.advisorRyan, Matthew
dc.contributor.advisorSkov, Peer
dc.contributor.authorDesilvestro, Adrian
dc.date.accessioned2020-04-14T23:57:29Z
dc.date.available2020-04-14T23:57:29Z
dc.date.copyright2020
dc.date.issued2020
dc.date.updated2020-04-09T07:40:35Z
dc.description.abstractProPublica’s allegations, that an algorithmic tool used to predict re-offenders is “biased against blacks”, met a wave of criticism from the wider community. Researchers have since shown a trade-off between accuracy and fairness, concluding that the risk tool, COMPAS, was not inherently discriminatory. However, in light of ProPublica’s objections, a growing body of literature on assessing fairness in machine learning systems has taken flight. Performance criteria combine quantitative and qualitative elements, so users ‘preferences’ are hard to specify objectively. This study explores a Pareto frontier framework to illustrate the relative model (in)efficiencies that arise in Risk Prediction Instruments (RPIs). The research follows a logistic framework for estimating recidivism risk, and the design parameters include the choice of fairness constraints and the choice of a bin scoring system (the “bin number”). This dissertation presents three experiments where decision-makers can improve performance in their RPIs: (1) improving efficiency through a relaxed version of the constraint, (2) improving efficiency through ‘cost-free’ constraint implementation, and (3) improving efficiency through a revised scoring system.en_NZ
dc.identifier.urihttps://hdl.handle.net/10292/13265
dc.language.isoenen_NZ
dc.publisherAuckland University of Technology
dc.rights.accessrightsOpenAccess
dc.subjectAlgorithmic fairnessen_NZ
dc.subjectAccuracy-fairness tradeoffsen_NZ
dc.subjectRisk prediction instrumentsen_NZ
dc.subjectPareto-frontier frameworken_NZ
dc.titleAbstraction and Prediction Algorithms: A Harm-Reduction Frameworken_NZ
dc.typeDissertationen_NZ
thesis.degree.grantorAuckland University of Technology
thesis.degree.levelMasters Dissertations
thesis.degree.nameMaster of Businessen_NZ
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
Dissertation [60pts] - Adrian Desilvestro.pdf
Size:
914.36 KB
Format:
Adobe Portable Document Format
Description:
Dissertation
Download
License bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
license.txt
Size:
895 B
Format:
Item-specific license agreed upon to submission
Description:
Download
Collections
Masters Dissertations