A Summary of Spatial Data Mining Softwares

 

Pusheng Zhang

University of Minnesota

 

I collected a list of popular softwares for spatial data mining, but it's far from a complete list. Since spatial data mining is such a new emerging field, many research prototypes and commercial softwares are being developed. The following softwares contain more or less spatial data mining functionalities, e.g., spatial classification, spatial clustering, spatial outlier detection, or spatial co-location mining. However, currently most spatial data mining researchers have to develop their own implementations due to the limited functionalities from available softwares.

 

(1) S-Plus Spatial Stats Module

Commercial software.

Webpage: http://www.insightful.com/products/product.asp?PID=17

 

Description:

S+SpatialStats is the first comprehensive, object-oriented software package for the analysis of spatial data. Providing a whole new set of analysis tools, S+SpatialStats was created specifically for the exploration and modeling of spatially correlated data.

 

Taking full advantage of the object-oriented methods and modeling language of S-PLUS, S+SpatialStats allows you to analyze data with spatial structure thoroughly and correctly. S+SpatialStats can be used to analyze data arising in technical areas such as environmental, mining and petroleum engineering, natural resources, geography, epidemiology, demography and others where data is sampled spatially.

 

Its major functionality includes the following:

 - Visualization/exploratory data analysis

 - Analyzing geostatistical data, e.g., kriging and variogram

 - Analyzing lattice data, e.g., spatial regression models

 - Analyzing spatial point patterns, e.g., K-function

 

(2) Spatial Modules for R

Free software

Webpage: http://sal.agecon.uiuc.edu/csiss/Rgeo/

Description:

There are several packages of spatial analysis available for R, almost all of which originated as S code and from quite different communities. As a result there is considerable overlap in their functionality. The following modules can assist spatial prediction(e.g., kriging) or spatial co-location pattern mining(e.g., cross-k function)

 

- spatial

Now part of the VR bundle. It contains trendsurface analysis, kriging and point-process code originally written by B. D. Ripley in 1990.

 

- splancs

Originally commercial code for spatial and space-time point patterns by Barry Rowlingson. Roger Bivand has made a GPL-ed version available for R, with contributions from Giovanni Petris.

 

- spatstat

Code for point pattern analysis(e.g., cross-k function) originally written for S-PLUS 5.1 by Adrian Baddeley and Rolf Turner. The version on the website (http://www.maths.uwa.edu.au/~adrian/spatstat.html) is said to work with R 1.0.1.

 

(3) Spatial Statistics Toolbox for Matlab/Fortran by K. Pace

Free software.

Matlab and Fortran toolbox for computing simultaneous and conditional spatial autoregressions and mixed regressive spatially autoregressive models, produced by K. Pace of the Dept. of Finance at Lousiana State University.

Website: http://www.spatial-statistics.com/

 

(4) Spatial Econometrics Library for Matlab

Free software

Webpage: http://www.spatial-econometrics.com/

Description: It's a full set of function expansion for spatial analysis, especially including spatial autoregressive modeling

 

(5) ClusterSeer/BoundarySeeer/SpaceStat by TeraSeer

Commercial software.

Webpage: http://www.terraseer.com

Description: The TeraSeer softwares support spatial clustering, spatial autocorrelation analysis, k-function, and classification.

 

(6) NEM

Free software

Description:

Neighborhood EM Clustering. This program clusters a given data set, using the Neighborhood EM algorithm proposed by Ambroise 1996.

Webpage: http://www.hds.utc.fr/~mdang/Progs/prognem.html

 

(7) GeoMiner

Description:

GeoMiner is a knowledge discovery system for spatial databases, being developed in the Database Systems Research Laboratory, School of Computing Science, Simon Fraser University. The data mining power of GeoMiner includes mining three kinds of rules: characteristic rules, comparison rules, and association rules, in geo-spatial databases, with a planned extension to include mining classification rules and clustering rules.

Webpage: http://db.cs.sfu.ca/GeoMiner/

 

(8) SPIN!

Webpage: http://www.ccg.leeds.ac.uk/spin/software.html

Description: The main objective of the SPIN! project is to offer new

possibilities for the analysis of georeferenced data. To this end a Spatial Data

Mining system is developed. It integrates state of the art Geographic Information

System and Data Mining functionality in an open, highly extensible,

internet-enabled plug-in architecture.

 

(9) GIS Softwares

The GIS softwares contain a full set of exploratory data analysis tools for spatial data, and these tools could be used to enhance spatial data mining. The following is a few major commercial GIS softwares:

 

 

ArcGIS

http://www.esri.com/software/arcgis/

 

Intergraph GeoMedia

http://imgs.intergraph.com/

 

MapInfo Professional

http://www.mapinfo.com/products/Overview.cfm?productid=1044