Our goal is to develop an open-source toolkit written in Python, which automates the analysis of performance and reliability for fleets of photovoltaic (PV) systems. All our algorithms use unlabled PV system production data, that is, measurements of real power production over time without reference meteorological data or system configuration information, which allows for the algorithms to work with the minimal possible data requirements. We primarly approach this work as a signal decomposition problem. For a list of publications related to this project please click here!