In this project, we aim to revolutionize the way circadian biologists study behavior. “Behavioral rhythms” are almost exclusively wheel-running rhythms, but these are only a subset of an animal’s behavioral repertoire. To overcome this, we use machine learning to automatically identify circadian rhythms in multiple behaviors simultaneously. We have used this method to measure sex- and estrogen-dependent differences in circadian behavior and are now investigating how rhythms change in the context of social interactions.