mukulu/ai-ethics-sdg-indicators

Resources for AI Ethics: Towards Ethics Aware Machine Learning Workflow - Innovating AI workflow with Ethics # AI Ethics and SDG Indicators: Data & Visualization Repository This repository contains the source codes, Jupyter Notebooks, and processed data files used in the analysis of Sustainable Development Goal (SDG) indicators, as presented in the conference paper: **"AI Ethics: Towards Ethics-Aware Machine Learning Workflow – Innovating AI Workflow with Ethics"** [Authors: John Francis Mukulu SJ, Sophia University, Tokyo, Japan] --- ## Overview This repository provides full transparency and reproducibility for all computational work discussed in the paper. It specifically focuses on mapping standard data science metrics to ethical frameworks (Utilitarianism, Virtue Ethics, Deontology) through the lens of SDG indicator analysis using **GDP per capita** data. The central code (Jupyter Notebook) and data files together demonstrate: - How classical ML metrics (mean, range, standard deviation) align with different ethical paradigms. - How different metrics can produce diverging ethical outcomes even on the same data. - Visualization techniques for uncovering and communicating these ethical divergences. --- ## Repository Structure . ├── GDP_per_capita_constant_2015_US.csv ├── filtered_gdp_data_1990_2022.csv ├── filtered_gdp_data_1990_2022_sovereign.csv ├── MetricsVisualizationVersion2.ipynb └── README.md --- ### File Descriptions - **GDP_per_capita_constant_2015_US.csv** Raw dataset: Historical GDP per capita (constant 2015 USD) for countries worldwide, 1990–2022. Source for all pre-processing and indicator calculation. - **filtered_gdp_data_1990_2022.csv** Processed dataset: Cleaned and filtered version, focusing on relevant countries and years. Used for intermediate analyses and visualizations. - **filtered_gdp_data_1990_2022_sovereign.csv** Processed dataset: Further filtered to include only sovereign states, as used in the main case study. Ensures direct comparability in all main figures. - **MetricsVisualizationVersion2.ipynb** Jupyter Notebook: The core cod …