Advisor: Tom Griffiths; Reader: Gabriel Vecchi
All are invited to attend. Please see abstract below.
Abstract:
This project investigates how people detect changes in continuous climate data (mean winter temperatures) and binary climate data (lake freeze). We find that participants shown the binary data detect greater change than participants shown the continuous data, even when correlation is controlled. We also show that using a Bayesian restrospective single changepoint detection algorithm on the data shows greater saliency of changepoints in binary data over continuous data, providing a possible explanation for the experimental results.