This paper studies the Random Utility Model (RUM) in a repeated stochastic choice situation, in which the decision maker is imperfectly informed about the payoffs associated to each of the available alternatives. By embedding the RUM into an online decision problem, we develop a gradient-based...