What is the Point of Change? Change Point Detection in Relational Event Models

Abstract

This paper presents an extension to the relational event model with change points (REM-CP) to study abrupt changes to social interaction behavior in temporal networks. A change point detection algorithm is proposed for exploring when and which network effects abruptly change, and a confirmatory approach to test the presence of a change point at a given moment. The effectiveness of the methodology was assessed with numerical simulations and NASA's Apollo 13 mission data. The latter revealed dynamic communication behavior and identified time zones where most change points occurred, including around the time of the famous quote “Houston, we've had a problem.”