An Efficient Formulation of Age Comparisons in the DISCRETE Edit System

The DISCRETE edit system, based on the Fellegi and Holt model [1976] of editing, contains two major components: edit generation and error localization. The set covering problem (SCP) is formulated with constraint matrices many times in both components. Therefore, an e cient set covering algorithm is critical to the overall performance of the DISCRETE edit system. The design of a set covering algorithm (Chen [1998]) provides a major performance improvement for the DISCRETE edit system. The size of the constraint matrices is a very important factor to have an e cient set covering algorithm. The age comparison approach described in Chen, Winkler, and Hemmig [2000] creates a huge number of new variables and edit rules in the set covering algorithm of the edit generation and error localization of the DISCRETE edit system. The dimension of the constraint matrices created is an increasing function of the number of variables and the number of edit rules. In this paper, we will describe an e cient formulation and simple implementation of the age comparisons in surveys that have the age eld.