Section 9 Fairness and Ethics

In this section, we dicuss some issues with fairness in prediction/classification, as well as broader ethical issues confronting the intersection of data science and social science.

Fairness in Machine Learning

Machine learning comes with a bundle of potential ethical issues. Hold on– we are doing machine learning? Yes, in our prediction/classification sections, we have done a type of it.

• We have used a statistical model (in this case, regression) to learn and make inferences about patterns in data.
  • A regression can be considered a type of algorithm
• We then apply the model to new data to make predictions and classify new data into categories.
• The models we have used are a type of “supervised machine learning” because our outcomes are pre-defined (vote share for Biden, a person donates vs. does not donate)
  • Other types of machine learning may be fully ``unsupervised,” such as searching data to come up with outputs, such as topics in books of text