[Openspace] A Question on Constructing Weights Matrix

Julia Koschinsky koschins at uiuc.edu
Wed Sep 14 10:28:12 CDT 2005


Michael,

You can construct a weights matrix in GeoDa for your data. 
1,500 observations won't be any problem. Here are the steps:

1. Download Geoda at 
https://www.geoda.uiuc.edu/downloadin.php and open the 
program

2. Go to Tools-Create-Weights and specify your input shape 
file, output file name, and unique ID

3. Decide what kind of neighborhood structure best fits your 
data and then choose a weights matrix format. For instance, 
if you have plots, you can choose a rook matrix if there are 
North-South, East-West effects (4 neighbors on a grid) or a 
queen matrix if you want to include your N-S, E-W neighbors 
as well as those sharing corners (8 neighbors on a grid).

4. Click Ok and you're done. If your shape file loads in 
GeoDa, you can make sure everything is fine with your 
weights matrix by opening it in the weights characteristics 
histogram.

5. You can use this weights matrix in Geoda or in other 
packages (e.g., in R, where you can manipulate/transform it 
further). You can open it in any text editor to get an 
overview of the format. The GeoDa manual at 
https://www.geoda.uiuc.edu/documentation.php#manuals 
provides additional detail.

Good luck,
Julia

---- Original message ----
>Date: Wed, 14 Sep 2005 09:13:20 -0600
>From: Michael Habteyonas <mzh at ualberta.ca>  
>Subject: [Openspace] A Question on Constructing Weights 
Matrix  
>To: openspace at sal.uiuc.edu
>
>I want to contstruct a W Matrix for my model.
>
>I have 1500 observations: plots of land 10miles by 10 miles 
in the province of 
>Alberta, Canada. Is this obervation too much to construct 
W? If it is not how 
>can I contruct W.
>
>I have not yet used any software, what is the best option 
to start with?
>
>Thanks for your help
>
>Michael 
>PhD Student
>Rural Economy
>UoA
>
>
>_______________________________________________
>Openspace mailing list
>Openspace at sal.uiuc.edu
>http://sal.uiuc.edu/mailman/listinfo/openspace


More information about the Openspace mailing list