Aggregation and heterogeneity of choice sets in discrete choice models

Aggregation methods for making aggregate predictions from disaggregate discrete choice models have previously been developed and analyzed by several authors. The representative agent, classification, and statistical differentials methods are modified to account for heterogeneity of choice sets. An empirical analysis that makes comparisons between the adjusted and unadjusted methods is also provided. The results indicate that the adjusted methods produce much more accurate aggregate predictions than the unadjusted methods.