Identification of hybrid systems is a well-researched topic, with most techniques relying on passive learning. Despite the abundance of data, active learning and counterexample-guided improvement of learned hybrid systems remain less explored. We present a passive and an active approach for the identification of state-dependent switched nonlinear systems with continuous state variables. We use segmentation in both of our approaches. Our passive approach consists of solving an optimization problem over a fixed set of data to find the continuous dynamics of the system-under-learning and the modes of training data, assuming a known number of modes, while our active approach incrementally learns these dynamics and mode information with no previous assumption on the number of modes. We showcase both of our approaches in a set of experiments, as well as a case study on electrical circuits.