Applying Measures of Spatial Autocorrelation Computation and Simulation推荐.pdfVIP

Geographical Analysis ISSN 0016-7363 Applying Measures of Spatial Autocorrelation: Computation and Simulation Roger Bivand Department of Economics, Norwegian School of Economics and Business Administration, Bergen, Norway Knowing about the challenges and opportunities of spatial autocorrelation is one thing, but applying the measures to one’s own data is another matter entirely. While manual computation of the measures for toy data sets is possible, applying them to small data sets required the use of computers and thus software. This article will shed some light on how the measures were and are implemented in software and on implementation issues that are still not fully resolved. Background When the empirical work reported in Cliff and Ord (1969) was being carried out, statistical computing was done by people in laboratory coats who kept researchers away from computing machinery. The Elliott 503 system used at that time at the University of Bristol had limited RAM (a fast section of 8 K words of 3911 bit length, and a backing section of up to 8 16 K word units, maximum 136 K words, 680 K bytes). Researchers interacted with the system by submitting code (in this case in Algol 60) and data for punching onto paper tape or cards. When computing jobs were prepared for executing, they were scheduled with those of other re- searchers, and turnaround times might be from hours to overnight—think of the time involved in ordering a library book from off-site storage. Just getting code to execute, let alone correctly represent formulae, was some- thing that took considerable resolve. In addition to the time delay of an execution cycle, memory was very limited, and very often researchers did not have access to standard libraries. Only in 1968 did Applied Statistics begin to publish algorithms, most often implemented