In this study, we investigate angle-of-arrival (AOA) and received signal strength (RSS) based localization methods for visible light communication (VLC) systems. We show that while AOA-based localization allows the receiver to locate itself via a least squares estimator by exploiting the directionality of light-emitting diodes (LEDs), RSS-based approach takes Lambertian pattern of LEDs into account and better deals with further improving the localization accuracy via a nonlinear least squares (NLS) estimator. In order to reduce the complexity of the NLS estimator, we develop an analytical learning rule based on the Newton-Raphson method and use the result of AOA-based localization as an initial point for the learning rule. As a benchmark, we also derive generic analytical expressions of the Cramer-Rao lower bound (CRLB) for RSS-based localization.