Spatial Data Analysis

Jack Bradbury (UCSD) and I have developed mac programs for the analysis of spatial data, particularly for use in animal behavior. Follow this link to his page. Recently, we have combined our efforts and first results are here. It was our goal to provide a general set of tools for an analysis of movement patterns and spatial distributions.

Here you find HomeRange 2.1.4 for PPC as a self-extracting archive with program and sample data files.

Documentation? What documentation? Here are the meager beginnings of a preliminary first draft - did I mention that it is not done yet?

This software is provided as postcard-ware (i.e., freeware). If you find it useful then send me a postcard at:

Robert Huber
Bowling Green State University, Biology, Bowling Green, OH 43403

The current version is version 2.1.4, released 13.8.99. Recent changes to the program are:

Version 2.0

• compiled for PowerPC and 68K: the analyses are a tad faster now :-)
• fixed several bugs since the beta release: eg. save as... works now properly if the file is saved under the same name

Version 2.0.1

• Convex/Minimum Polygon fixed bug when captures are on one line
• Monte Carlo Simulations: now uses a random seed when the seed field is left blank
• Changes in the underlying TCL structure: the 3D modelling has been moved to a separate class of document
• Tesselations: Streamlined code for improved speed

Version 2.1

• updated code for the TCL 1.3, improved memory handling, fixed bugs in the preloads routine
• Tesselations: fixed some bugs and improved memory handling

Version 2.1.4

• random number geberator: improved the random number generator in the Monte Carlo Simulation

The following things still need improving in the current version (thanks to Tom Weiss and others for suggestions, criticisms, help):

• Harmonic Mean: all measures and all graphs using the calculations of harmonic means are not quite right
• 3D surfaces: UDs from bivariate normal and normal distributions are a bit off.
• Printing data files: only part of each datafile page is imaged to the printer
• Spatial Autocorrelation: gets "in a funk" when the number of grid cells gets large (>10 for rows and columns)

Planned features in future releases - actually - this is now on hold - Metrowerks has just decided to discontinue support for Pascal:

• add standard normal deviates for spatial autocorrelation, right now only Morans i is supported
• plotless stats and fractals: not yet implemented
• add "restricted edge" to polygon calculations
• add GIS data file support for in/output (eg, Arc/info, IDRISI, GRASS)
• add Contiguous Clustering: I got the source code, I just need to stick it in and smooth over the interface

Also check out link collections for analysis software in biology, wildlife ecology, and ethology

Check out the Fieldworker for interfacing GPS systems with the Apple Newton. This offers a wonderful new way to record localities and the spatial distribution of various characteristics.

Lobsterman's Programming

I have been programming since 1980, first in FORTRAN on the DEC Vax, some on IBM compatibles, and then since 85 for the Apple Macintosh (I used the Apple Lisa in those early days). My programming is mostly done in Pascal, some in - uuugh - C/C++, and most of my recent work uses Java.

Many of my projects to date have been based on the Pascal Think Class Library framework (Symantec), compiled with the Think Pascal (Symantec) compiler. Symantec has unfortunately turned into useless mush and the library has been ported to Codewarrior by some kind souls. I have since ported my own projects to the Metrowerks Codewarrior environment. Another Pascal option is based on the MacApp object library and it is kept current at MacApp2PPC.

I then installed the BeOS on one of my Macs and was quite impressed by its features. I bought a Newton Messagepad 130 and did a bit of coding for both of these platforms. Now that the Newton is dead and the BeOS is a bit less tempting with the revival of the Macintosh, I have decided to stay a bit longer with the platform that has given me pleasure and good vibes for more than a decade.

My main coding efforts at this point are going into Java. Here is my site with Java programs useful for analyses in animal behavior - check out what I cooked up so far: Java Grinders.

• lobsterman's homePage
• lobsterman's object libraries
• lobsterman's Java applets
• lobsterman's programming