[GRASS-stats] Cross-checking a custom function for separability indices

Nikos Alexandris nikos.alexandris at felis.uni-freiburg.de
Wed Apr 21 21:05:31 EDT 2010


Hi list.

This is not exactly a "grass-"stats topic but I require help to cross-compare 
results from a custom R function I wrote. Since many spatial analysts read 
this list, I guess that there is a good chance that someone experienced can 
help me out.

I calculate via  a function various separability indices or distances for 
samples between classes, namely: Bhattacharrya, Jeffries-Matusita, Divergence 
and Transformed Divergence (...it's true that I have posted about the 
Bhattacharyya distance in the past in this list [2]).

I need to verify the reliability of the function. I don't know of any R 
package or other tool to get those measures quickly calculated. The one (foss-
tool) I found that contains a function to calculate the Divergence is the LAS 
[2]. But I have trouble getting it compiled under bash.

The function and a call for help is posted already in  r-help _at_ r-
project.org [3]. Does anybody have access to any tool that could quickly 
calculate those indices (actually just the Divergence and the Bhattacharryya 
would be enough)?

Thank you, Nikos


--%<---
Two small samples (in R code style --- the same posted in r-help):

# two samples
sample.1 <- c (1362, 1411, 1457, 1735, 1621, 1621, 1791, 1863, 1863, 1838)
sample.2 <- c (1354, 1458, 1458, 1458, 1550, 1145, 1428, 1573, 1573, 1657)

# running the custom function (posted in r-help)
separability.measures ( sample.1 , sample.2 )

Divergence:             1.5658 
Bhattacharryya:         0.1683 
Jeffries-Matusita:      0.3098 
Transformed divergence: 0.3555


--%<---
[1] http://osgeo-org.1803224.n2.nabble.com/How-to-calculate-the-Bhattacharyya-
distance-index-td2317804.html

[2] ftp://dbftp.essc.psu.edu/pub/code/las

[3] http://n4.nabble.com/Cross-checking-a-custom-function-for-separability-
indices-td2019342.html#a2019342


More information about the grass-stats mailing list