Filling in Missing Values in the North Dakota Land Valuation Model

Abstract

The North Dakota Land Valuation Model relies on a crop yield data set published by the USDA's National Agricultural Statistics Service. Over the years this data set, due to the way in which the data is collected, has developed a growing problem with missing values. This research utilizes a secondary data set and the spatial nature of the crop yield data to develop a new methodology for interpolating the missing values in the NASS data set. This methodology uses the secondary data set to establish the spatial nature of the yield data and to build spatial weights matrices. These spatial weights matrices are used in conjunction with the secondary data set and weather data to develop various methods of estimating the missing values. Once this is done the results of the methodologies are analyzed for both individual and global convergence.