Learning and Earning Under Noise and Uncertainty

Sequential decision-making under uncertainty is central to a range of operations and marketing problems. In the face of an unknown environment, decision-maker needs to strike a balance between learning the known environment (“learning”) and selecting nearly optimal decisions (“earning”). For instance, consider pricing a new product. If the retailer had full information about the demand at every price level, then she could determine the revenue-maximizing price for the good. However, such information about the demand curve is typically not available in practice, so the seller needs to experiment with different prices to gain information about the demand curve, and then exploit this information to offer a near-optimal selling price.