In this paper, we provide a novel framework for decision making under uncertainty based on information available in the form of a data set of cases. A case contains information about an action taken, an outcomes obtained, and other circumstances that were recorded with the action and the outcome. The set of actions, the set of outcomes and the set of possibly relevant recorded characteristics are derived from the cases in the data set. The information from the data set induces a belief function over outcomes for each action. From a decision maker's preferences over belief functions one can derive a representation evaluating outcomes according to the α-max min criterion. New data affects behavioral parameters, such as awareness, ambiguity and ambiguity attitude, and may suggest a classifications of data into states.