Scalable Parallel Approximate Formulations of Multidimensional Spatial Auto-Regression Models for Spatial Data Mining

摘要

The spatial auto-regression (SAR) model is a popular spatial data analysis technique which has been used in many applications with geo-spatial datasets. However, exact solutions for estimating SAR parameters are computationally expensive due to the need to compute all the eigen-values of a very large matrix. Therefore, serial solutions for the SAR model do not scale up to map sizes of interest to the Army. Thus, we developed the parallel approximate SAR models which can now be used by the Army to increase the accuracy and usefulness of maps, better analyze the impact of weather on the battlefield, make near-future predictions of the locations of enemy units, and increase the lethality of missiles.