By Andrea Lane (Duke University)
Information
Many of us have used some version of the Titanic dataset for logistic regression. I avoided it for a long time because it was so common—until my own interest in the Titanic tragedy inspired me to create an immersive activity for interpreting logistic regression coefficients. In this activity, each student assumes the identity of a real Titanic passenger, selecting from a list generated with the help of AI. They learn details about their passenger, including class, age, and gender. Using these features, we fit a logistic regression model predicting survival. Students then calculate their own probability of survival and compare it to the actual outcome. Along the way, I share fun facts about the Titanic, and to make the experience more engaging, My Heart Will Go On plays while students work on a brief paired exercise. Completed worksheets, which include passenger details, predicted survival probability, and actual outcome, are graded on completion to encourage participation. This activity is designed for a statistical modeling course for first-year data science master’s students, with a class size of 45, though it can be scaled up or down. The activity directly connects to sparking joy and engagement in students. By becoming part of the Titanic story, they experience the data in a personal and memorable way, while actively applying logistic regression concepts. The use of AI to generate passenger details adds authenticity and variety to the activity, making it both fun and educational.