CONSTRUCTING AND AGGREGATING DATABASES.

This paper presents a technique for constructing a sample database to represent a testable population with known characteristics. An optimization model is proposed for solving this problem. The resulting model is amenable to solution by a heuristic-subgradient algorithm. Computational results are discussed along with descriptions of two practical applications of the methdology: (1) reducing the statistics of the income taxpaper file for the U. S. Treasury Department, and (2) forming an index fund of common stocks. The latter example illustrates a dynamic environment.