Roger Bivand
Thu, 15 Apr 2004 10:41:23 -0700
On Thu, 15 Apr 2004, Munroe, Darla K wrote: > From experience, I am somewhat suspicious of the Delaunay triangulation > routine in the Matlab Spatial Econometrics Toolbox. I'm tried to replicate > matrices that I calculated with the simplest of datasets (i.e., a raster > grid), and not gotten the same results. If you are using Matlab to do your > modeling, it's definitely the best that's out there, but there are some > other tools available.
Well, like most triangulation algorithms, it does not know how to choose between four (raster grid) equidistant points, so probably numerical fuzz in the coordinate positions will give different neighbours each time. Triangulation works for irregularly spaced points, like irregular polygon centroids, and typically will be close to the weights matrix for contiguous boundaries too. So I think the Matlab code does what it claims to do, but it can't do the impossible. For raster cells we know the neighbours anyway, don't we? > > In particular, there is Luc Anselin's GeoDa package, which is free. It's > still in beta format, but the final release will be sometime this fall. > It's really a wonderful package for all sorts of data exploration and > spatial data exploration. It can handle tens of thousands of obs (which > SpaceStat couldn't), and it can create all sorts of weights matrix, as well > as estimate a couple spatial regression models: > > http://sal.agecon.uiuc.edu/geoda_main.php > The fresh spdep package for R (0.2-14) due on CRAN any time now has functions for reading and writing GeoDa GAL and GWT files and Matlab *.dat files for weights, as well as tools for generating, editing, subsetting, and summarising neighbour and weight lists. In response to the original question, I think that you will find that there are a number of variants of the Columbus data set - the coordinates in the one in spdep for R have an x range 6-11.5, y range 10.5-15. I'm not sure if they are projected, or what projection - nor is ArcMap - these are the same boundaries as come with GeoDa in files COLUMBUS.shp and col2.shp. I believe colleagues at Ohio State have some idea - I've done this once before to attach GPS points with pictures to the data, and your measured coordinates are correct. Contact me off-list if you like. Roger Bivand > - DM > > > > -----Original Message----- > From: Mark Coleman [mailto:[EMAIL PROTECTED] > Sent: Thursday, April 15, 2004 9:42 AM > To: [EMAIL PROTECTED] > Subject: AI-GEOSTATS: "Naive" question on lat-long conventions > > > Greetings, > > I apologize in advance for what must be a rather naive question, but I > am new to the field of spatial statistics. In the course of learning > and testing various estimation methods with test data sets, I have a > couple of questions regarding conventions regarding lat-long > coordinates. In particular, I have been using a small test data set > provided by Anselin on crime data in Columbus OH. Each sample point > includes a lat-long coordinate. The data set can be found at > http://www.spatial-econometrics.com/data/ > > As an example, the first point in this data set shows latitude=35.62, > longitude=42.38. Yet when I check the coordinates of a random point in > Columbus OH (not in the dataset), I find coordinates in the range of > Lat=40.x, Long= -82.y. I use the points in the Anselin data set to > calculate a spatial distance matrix using Delaunay Triangulization. I > wanted to be sure that I understand what the conventions are when > encoding lat-long coordinates. In the end, I need to understand this so > I can calculate the distance between two points based on their lat-long > coordinates. > > Thanks, > > Mark > > > -- > * To post a message to the list, send it to [EMAIL PROTECTED] > * As a general service to the users, please remember to post a summary of > any useful responses to your questions. > * To unsubscribe, send an email to [EMAIL PROTECTED] with no subject and > "unsubscribe ai-geostats" followed by "end" on the next line in the message > body. DO NOT SEND Subscribe/Unsubscribe requests to the list > * Support to the list is provided at http://www.ai-geostats.org > > -- > * To post a message to the list, send it to [EMAIL PROTECTED] > * As a general service to the users, please remember to post a summary of any useful > responses to your questions. > * To unsubscribe, send an email to [EMAIL PROTECTED] with no subject and > "unsubscribe ai-geostats" followed by "end" on the next line in the message body. DO > NOT SEND Subscribe/Unsubscribe requests to the list > * Support to the list is provided at http://www.ai-geostats.org > -- Roger Bivand Economic Geography Section, Department of Economics, Norwegian School of Economics and Business Administration, Breiviksveien 40, N-5045 Bergen, Norway. voice: +47 55 95 93 55; fax +47 55 95 93 93 e-mail: [EMAIL PROTECTED] -- * To post a message to the list, send it to [EMAIL PROTECTED] * As a general service to the users, please remember to post a summary of any useful responses to your questions. * To unsubscribe, send an email to [EMAIL PROTECTED] with no subject and "unsubscribe ai-geostats" followed by "end" on the next line in the message body. DO NOT SEND Subscribe/Unsubscribe requests to the list * Support to the list is provided at http://www.ai-geostats.org