Variable Importance
Standard and conditional variable importance for ‘cforest’, following the permutation principle of the ‘mean decrease in accuracy’ importance in ‘randomForest’.
varimp(object, mincriterion = 0, conditional = FALSE,
threshold = 0.2, nperm = 1, OOB = TRUE, pre1.0_0 = conditional)
varimpAUC(...)object |
an object as returned by |
mincriterion |
the value of the test statistic or 1 - p-value that
must be exceeded in order to include a split in the
computation of the importance. The default |
conditional |
a logical determining whether unconditional or conditional computation of the importance is performed. |
threshold |
the threshold value for (1 - p-value) of the association
between the variable of interest and a covariate, which must be
exceeded inorder to include the covariate in the conditioning
scheme for the variable of interest (only relevant if
|
nperm |
the number of permutations performed. |
OOB |
a logical determining whether the importance is computed from the out-of-bag sample or the learning sample (not suggested). |
pre1.0_0 |
Prior to party version 1.0-0, the actual data values were permuted according to the original permutation importance suggested by Breiman (2001). Now the assignments to child nodes of splits in the variable of interest are permuted as described by Hapfelmeier et al. (2012), which allows for missing values in the explanatory variables and is more efficient wrt memory consumption and computing time. This method does not apply to conditional variable importances. |
... |
Arguments to |
Function varimp can be used to compute variable importance measures
similar to those computed by importance. Besides the
standard version, a conditional version is available, that adjusts for correlations between
predictor variables.
If conditional = TRUE, the importance of each variable is computed by permuting
within a grid defined by the covariates that are associated (with 1 - p-value
greater than threshold) to the variable of interest.
The resulting variable importance score is conditional in the sense of beta coefficients in
regression models, but represents the effect of a variable in both main effects and interactions.
See Strobl et al. (2008) for details.
Note, however, that all random forest results are subject to random variation. Thus, before
interpreting the importance ranking, check whether the same ranking is achieved with a
different random seed – or otherwise increase the number of trees ntree in
ctree_control.
Note that in the presence of missings in the predictor variables the procedure described in Hapfelmeier et al. (2012) is performed.
Function varimpAUC is a wrapper for
varImpAUC which implements AUC-based variables importances as
described by Janitza et al. (2012). Here, the area under the curve
instead of the accuracy is used to calculate the importance of each variable.
This AUC-based variable importance measure is more robust towards class imbalance.
For right-censored responses, varimp uses the integrated Brier score as a
risk measure for computing variable importances. This feature is extremely slow and
experimental; use at your own risk.
A vector of ‘mean decrease in accuracy’ importance scores.
Leo Breiman (2001). Random Forests. Machine Learning, 45(1), 5–32.
Alexander Hapfelmeier, Torsten Hothorn, Kurt Ulm, and Carolin Strobl (2012). A New Variable Importance Measure for Random Forests with Missing Data. Statistics and Computing, doi: 10.1007/s11222-012-9349-1
Torsten Hothorn, Kurt Hornik, and Achim Zeileis (2006b). Unbiased Recursive Partitioning: A Conditional Inference Framework. Journal of Computational and Graphical Statistics, 15 (3), 651-674. Preprint available from https://www.zeileis.org/papers/Hothorn+Hornik+Zeileis-2006.pdf
Silke Janitza, Carolin Strobl and Anne-Laure Boulesteix (2013). An AUC-based Permutation Variable Importance Measure for Random Forests. BMC Bioinformatics.2013, 14 119. doi: 10.1186/1471-2105-14-119
Carolin Strobl, Anne-Laure Boulesteix, Thomas Kneib, Thomas Augustin, and Achim Zeileis (2008). Conditional Variable Importance for Random Forests. BMC Bioinformatics, 9, 307. doi: 10.1186/1471-2105-9-307
set.seed(290875)
readingSkills.cf <- cforest(score ~ ., data = readingSkills,
control = cforest_unbiased(mtry = 2, ntree = 50))
# standard importance
varimp(readingSkills.cf)
# the same modulo random variation
varimp(readingSkills.cf, pre1.0_0 = TRUE)
# conditional importance, may take a while...
varimp(readingSkills.cf, conditional = TRUE)
## Not run:
data("GBSG2", package = "TH.data")
### add a random covariate for sanity check
set.seed(29)
GBSG2$rand <- runif(nrow(GBSG2))
object <- cforest(Surv(time, cens) ~ ., data = GBSG2,
control = cforest_unbiased(ntree = 20))
vi <- varimp(object)
### compare variable importances and absolute z-statistics
layout(matrix(1:2))
barplot(vi)
barplot(abs(summary(coxph(Surv(time, cens) ~ ., data = GBSG2))$coeff[,"z"]))
### looks more or less the same
## End(Not run)Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.