Comment on “A Conditional Dependence Adjusted Weights of Evidence Model” by Minfeng Deng in Natural Resources Research 18(2009), 249–258
- 132 Downloads
The major subject of our comment is the “formal expression for the bias of the contrasts” of weights of evidence (Deng 2009, p. 249), in particular Equation (15) and its generalization to Equation (16) (Deng 2009, p. 252), and the claim that “a modified WE (weights of evidence, these authors) model will then be developed, where the bias is corrected using the correlation structure of the predictor patterns” (Deng 2009, p. 249) and the conclusion that it “is on par with LR (logistic regression, these authors), which is not affected by violation of CI (conditional independence, these authors)” (Deng 2009, p. 257). We will show by way of a counterexample that Equation (16) of Deng (2009, p. 252) does not generally hold and clarify that its derivation is flawed by mistaking that the logit transform is linear, i.e. that logit and summation commutate.
KeywordsLogistic Regression Conditional Independence Simple Linear Relationship Evidence Model Unique Notation
Instructive discussions on marginal odds with Prof. Peter E. Jupp, School of Mathematics and Statistics, University of St. Andrews, Scotland, are gratefully acknowledged.
- Agterberg, F. P. (2011). A modified weights-of-evidence method for regional mineral resource estimation. Natural Resources Research. doi: 10.1007/s11053-011-9138-0.
- Agterberg, F. P., & Bonham-Carter, G. F. (1999). Logistics regression and weights of evidence modeling in mineral exploration. In Proc. APCOM’99, computer applications in the minerals industries (pp. 583–590). Colorado School of Mines.Google Scholar
- Agterberg, F. P., Bonham-Carter, G. F., Wright, D. F., & Cheng, Q. (1989). Weights of evidence and weighted logistic regression for mineral potential mapping. In J. C. Davis & U. C. Herzfeld (Eds.), Computers in geology: 25 years of progress (pp. 13–32). New York: Oxford University Press.Google Scholar
- R Development Core Team. (2008). R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing. ISBN 3-900051-07-0. http://www.R-project.or.