Reproduction of: Retiring Adult - New Datasets for Fair Machine Learning (IS477)

This repository is based on the results from the paper, "Retiring Adult: New Datasets for Fair Machine Learning" by Frances Ding, Moritz Hardt, John Miller and Ludwig Schmidt. In this assignment, our primary goal is to replicate specific numerical outcomes by employing a straightforward logistic regression model. The focus of our replication efforts is centered on achieving results as follows: First, we aim to reproduce the overall accuracy rate of 85% achieved by a standard logistic regression model trained on the UCI Adult dataset. Additionally, we seek to duplicate the model's performance in terms of accuracy, attaining 91.4% for instances categorized as "Black" and 92.7% for instances classified as "Female." These replication endeavors are crucial in assessing the reproducibility and consistency of the initial results, contributing to the integrity and reliability of our research findings. This repository and the associated project aim to provide essential skills and knowledge in data science. Core objectives include effective data management, legal and ethical awareness, selection of appropriate licenses, fostering reproducibility and transparency, understanding privacy and legal considerations, data quality management, workflow automation, recognizing the significance of metadata, long-term archiving, and proper citation practices. The project utilizes the UCI Adult dataset, known for its sensitivity attributes, to practically apply these principles and skills.