We provide a methodology for estimating the losses due to shade in power generation data sets produced by real-world photovoltaic (PV) systems. We focus this work on estimating shade loss from data that are unlabeled, i.e. power measurements with time stamps but no other information such as site configuration or meteorological data. This approach enables, for the first time, the analysis of data generated by small scale, distributed PV systems, which do not have the data quality or richness of large, utility-scale PV systems or research-grade installations. This work is an application of the newly published signal decomposition (SD) framework, which provides an extensible approach for estimating hidden components in time-series data.