Premium
Redundancy as an Index of Change in Point Pattern Analysis
Author(s) -
Lews Ralph
Publication year - 1979
Publication title -
geographical analysis
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.773
H-Index - 65
eISSN - 1538-4632
pISSN - 0016-7363
DOI - 10.1111/j.1538-4632.1979.tb00703.x
Subject(s) - poisson distribution , entropy (arrow of time) , redundancy (engineering) , mathematics , polygon (computer graphics) , point pattern analysis , principle of maximum entropy , statistics , algorithm , computer science , information theory , pattern recognition (psychology) , data mining , common spatial pattern , artificial intelligence , telecommunications , physics , quantum mechanics , frame (networking) , operating system
Point pattern analysis based on concepts from information theory can go beyond existing techniques. Direct measurement of spatial form is achieved when Thiessen polygons are constructed around the points; in this scheme, each point's proportion of total area may be treated like a probability. Three information‐theoretic indices are available for analysis of a distribution of such probabilities. Entropy is density dependent. Redundancy, defined as the difference between observed and maximum entropy, seems to avoid this problem when the number of individuals in a pattern exceeds twenty. Comparisons of prior and posterior redundancy provide an indication of change in overall pattern form. An information gain expectation reflects changes for each individual in a pattern. Here, point‐area redundancy parameters are determined for Poisson‐generated patterns, using a gamma distribution of polygon areas and a computer‐generated set. An application to an urban crime problem illustrates the use of these parameters in the analysis of pattern change .