Using Copula Methods to Create Synthetic Data Sets for Agricultural Applications

• Activity Number: 229
• Type: Topic Contributed
• Date/Time: Monday, August 2, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Timothy Paul Keller* and Darcy Miller+
• Companies: National Agricultural Statistics Service and National Agricultural Statistics Service
• Keywords: copulas ; agriculture ; simulation ; data augmentation

Abstract

Creating synthetic copies of a data set is useful for many reasons: particularly,being able to conduct simulations that overcome the limitations posed by a small data set,and addressing concerns in external data sharing. A synthetic copy of a real data set suitable for analysis preserves the marginal distributions of each variable and their joint behavior. Designing a general purpose method requires the estimation of the marginals to be data driven. Copulas are functions that combine univariate distributions to form a joint distribution with a specified dependence structure. Applications of copulas span the fields of business,actuarial science,engineering,biomedicine,and spatial science. The National Agricultural Statistics Service is investigating data driven estimation of marginals and using copulas to augment or simulate data sets with the integrity for valid statistical analysis.