Kim Murray and Mary M. Conner. 2009. Methods to quantify variable importance: implications for the analysis of noisy ecological data. Ecology 90:348–355.

Supplement

Source code for variable importance situations.
Ecological Archives E090-026-S1.

Copyright

Authors
File list (downloads)
Description

Author(s)

Kim Murray Berger
205 Natural Science Building
University of Montana
Missoula, MT 59812 USA
E-mail: kim@snowleopard.org

Mary M. Conner
Department of Wildland Resources
Utah State University
Logan, UT 84322-5230 USA

File list

Variable Importance Simulation.txt

Variable Importance Simulation.sas

hierpart.txt

hierpart.sas

Description

"Variable Importance Simulation.sas" is a simulation to evaluate the relative importance of random variables using Akaike weights, standardized regression coefficients, partial and semi-partial correlation coefficients, and hierarchical partitioning. Remember to change the subdirectory where hierpart.sas is called from the include statement.

The file "hierpart.sas" is a macro that executes hierarchical partitioning analysis as described by Chevan and Sutherland in American Statistician, 1991, Vol. 45, no. 2, pp. 90–96. This macro was written by Kim Murray Berger and Mary M. Conner based on a Dominance Analysis macro written by Razia Azen and Robert Ceurvorst (http://www.uwm.edu/~azen/damacro.html). Hierarchical Partitioning analysis quantifies the importance of each predictor as its average contribution to the model r-square, across all possible models. Note: This program is limited to at most 10 predictors!