Data mining techniques have been used on data collected from a photovoltaic system to predict its generation and performance. Nevertheless, up to date, this computing approach has needed the simultaneous measurement of environmental parameters that are collected by an array of sensors. This chapter presents the application of several computing learning techniques to electrical data in order to detect and classify the occurrence of failures (i.e. shadows, bad weather conditions, etc.) without using environmental data. The results of a 222kWp (CdTe) case study show how the application of computing learning algorithms can be used to improve the management and performance of photovoltaic generators without relying on environmental parameters.