Advanced utility meters, which have the capability to measure and transmit high resolution usage data in real time, are growing in popularity. A switch from traditional meters to smart meters can reduce labor costs for meter reading and increase customer engagement, but the full potential of this technology often remains untapped. In this thesis, we demonstrate one procedure for deriving more meaningful information from the high resolution and real-time data that these devices can provide. Using hourly consumption data for water, electricity, steam and chilled water from three buildings at the University of Texas at Austin, we superimposed synthetic “events”, or abnormal periods of consumption, and then developed a system to detect these events. This detection is achieved by first fitting predictive models to the unmodified data, and then analyzing the model residuals on new data to mark a data point as unusual if the deviation between the observed and predicted value exceeds a certain threshold