Data Science Ethics and Governance
As data science impacts decisions affecting lives, ethical responsibility and governance are essential.
Responsible Data Science
Fairness (no discrimination), accountability (someone responsible), transparency (understandable decisions), ethics (broader societal impact).
Bias
Sampling bias, historical bias, measurement bias, algorithm bias. Mitigation: diverse data, bias testing, fairness constraints, regular auditing.
Privacy
Data minimisation, anonymisation, differential privacy, consent. Comply with GDPR, CCPA, local privacy laws.
Data Governance
Policies for data quality, security, access, usage. Data catalogues, lineage, quality rules, access controls, retention policies.
Model Governance
Development standards, validation, production monitoring, version control, documentation. Model cards document purpose, performance, limitations.
Regulatory Landscape
GDPR (right to explanation), EU AI Act (risk-based), CCPA (consumer rights). Nepal developing data protection legislation.
Summary
Data science ethics and governance ensure responsible use of data and models through bias mitigation, privacy, and compliance.