A common problem in estimation of discrete choice models is that the complete choice set is very large.  A good example is supermarket consumer goods, like breakfast cereal where there are often a hundred or more varieties (SKUs or UPCs) to choose from.  In that case, estimation of complex...