Explainable Machine Learning for Transparent Public-Service Decision Systems

Lucia Bianchi

Explainable Machine Learning for Transparent Public-Service Decision Systems

Author(s)	Dr. Lucia Bianchi
Country	United States
Abstract	Machine learning is increasingly relevant to public-service administration, where predictive systems can assist with case prioritization, resource allocation, application screening, complaint routing, inspection planning, and service-delivery management. However, public-sector decisions differ from many commercial predictions because they may influence access to public benefits, administrative opportunities, essential services, or regulatory attention. Predictive accuracy alone is therefore insufficient when citizens and public officials cannot understand the principal factors influencing a recommendation. This study proposes an Explainable Public-Service Machine Learning Framework designed to combine predictive performance with transparent reasoning, reviewability, human oversight, and auditable decision records. The proposed architecture distinguishes between technical model interpretation, officer-facing decision explanations, and citizen-facing reason communication. It incorporates interpretable modeling where feasible, feature-attribution analysis for complex models, confidence disclosure, standardized reason codes, lawful counterfactual information, human escalation, and decision logging. A design-science methodology was combined with structured evidence mapping and a reproducible scenario-based simulation. A total of 2,000 synthetic decision episodes were distributed across five public-service domains: social assistance, permit prioritization, grievance triage, housing maintenance, and public-health inspection. An opaque machine-learning configuration was compared with the proposed explainable configuration using decision accuracy, a transparency index, and administrative review time. Simulation results showed comparable predictive accuracy, with 79.5% for the opaque configuration and 78.3% for the explainable configuration. In contrast, the mean transparency index increased from 43.3% to 90.9%, while modeled review time declined from 12.0 to 7.4 minutes. These results illustrate that substantial gains in decision transparency and review efficiency can be achieved without requiring large reductions in predictive performance. The study concludes that public-service machine learning should be evaluated as an accountable decision-support infrastructure rather than solely as a predictive algorithm. Transparent public administration requires explanations that are technically meaningful, understandable to affected persons, reviewable by officials, and sufficiently documented to support correction, appeal, and institutional accountability.
Keywords	explainable machine learning; public-service decision systems; algorithmic transparency; interpretable machine learning; public administration; human oversight; decision accountability; SHAP; transparent governance; responsible artificial intelligence
Field	Engineering
Published In	Volume 1, Issue 1, January-February 2020
Published On	2020-01-09

View / Download PDF File

About AJAMIR Fees & Payment Current Issue Publication Archive	Submit Research Paper Track Submission Status Publication Guidelines Publication Ethics Peer Review & Plagiarism	Join as a Reviewer Editors & Reviewers Reviewer Referral Program Get Reviewer Membership Certi.	Website/Journal Policies Usage Policy Content Policies Privacy Policy

Contact Us		+1-872-302-3851	editor@ajamir.com

American Journal of Advanced Multidisciplinary Innovation and Research

A Widely Indexed Open Access Peer Reviewed Multidisciplinary Bi-monthly Scholarly International Journal

Explainable Machine Learning for Transparent Public-Service Decision Systems

Share this