
Fundamental Review on the Formulation of Large Lattice Spatial Neighbor MatricesFundamental Review on the Formulation of Large Lattice Spatial Neighbor MatricesAN00408380 Fundamental Review on the Formulation of Large Lattics Spatial Neighbor Matrices 
"/Fitrianto,Gigih/"Fitrianto,Gigih ,
"/田中,章司郎/"田中,章司郎
40
(
3
)
, pp.41

88 , 20171231 , 広島経済大学経済学会
ISSN:03871444
NCID:AN00408380
Description
To calculate the impact of each location within an observation area we need to calculate the dependences of that location. In order to specify the model to explain this condition, we must define the neighbor relation for each location. This important information is described by a spatial neighbor matrix (Cressie, 1991: Ch. 6). By using Spatial Matrix Dr×c, which is extracted from polygon structure of spatial lattice DM, we can construct Neighbor Relation Matrix, W. There should be several methods to construct W matrix, such as: 1) Direct Arrow Reading (DAR); 2) InnerOuter Neighbor Matrix (ION); and 3) Kronecker Product.
In this research, we verified the algorithm performances based on their time and space efficiency. All of them were calculated based on the complexity and real execution. We found that Kronecker product method became the best method to construct W matrix. That method can be used efficiently both in terms of computational time and space.
FullText
http://harp.lib.hiroshimau.ac.jp/hue/file/12365/20180312155306/kenkyu2017400303.pdf