On March 4, 2019, the DADD team was joined by José Ortega from BigML for a workshop on machine learning and digital discrimination.
BigML offers a web-based Machine Learning platform that includes a selection of robustly-engineered Machine Learning algorithms, both supervised and unsupervised, such as classification and regression, cluster analysis, anomaly detection, and topic modeling.
Students at the Department of Digital Humanities worked with BigML’s platform to discover discriminatory patterns in different, real-world datasets. They also presented their findings in one-minute presentations. The key takeaway of the day: datasets rarely are obviously discriminatory. Pruning the dataset and tracing proxy variables are key practices!