A Study on Imputation for Missing Data using the Kriging

Geostatistics is a branch of statistical theory concerned with problems of spatial serial data, interpolation and mapping of distributed data, and related problems. The missing data are produced in many fields concerned with geostatistical data. Generally, in case of continuous data, we use the average as a method for imputation of missing data. However, using an average is no longer a good method to impute for the geostatistical missing data as there exist a spatial autocorrelation among them. In this study, we suggest the ‘kriging’ that is spatial statistics method as a method to impute for those spatial missing data, and then compare it with general average method in aspect to imputation superiority. To show the usefulness of the proposed method, we perform a small simulation study based on the PRESS statistic and show an empirical example with rainfall data from the Philippines.