Estimate variable importances in an earth object
Estimate variable importances in an
evimp(object, trim=TRUE, sqrt.=TRUE)
TRUE(default), delete rows in the returned matrix for variables that don't appear in any subsets.
TRUE, meaning take the
sqrtof the GCV and RSS importances before normalizing to 0 to 100. Taking the square root gives a better indication of relative importances because the raw importances are calculated using a sum of squares. Use
FALSEto not take the square root.
This function returns a matrix showing the relative importances of the
variables in the model. There is a row for each variable. The row
name is the variable name, but with
col: Column index of the variable in the
used: 1 if the variable is used in the final model, else 0. Equivalently, 0 if the row name has an
nsubsets: Variable importance using the "number of subsets" criterion. Is the number of subsets that include the variable (see "Three Criteria" in the chapter on
earthvignette Notes on the earth package).
gcv: Variable importance using the GCV criterion (see "Three Criteria").
gcv.match: 1, except is 0 where the rank using the
gcvcriterion differs from that using the
nsubsetscriterion. In other words, there is a 0 for values that increase as you go down the
rss: Variable importance using the RSS criterion (see "Three Criteria").
gcv.matchbut for the
-unusedappended if the variable does not appear in the final model.The columns of the matrix are (not all of these are printed by
nsubsetscriterion. This means that values in the
nsubsetscolumn decrease as you go down the column (more accurately, they are non-increasing). The values in the
rsscolumns are also non-increasing, except where the
rssrank differs from the
There is a chapter on
evimp in the
earth package vignette
Notes on the earth package.
Thanks to Max Kuhn for the original
evimp code and for helpful discussions.
data(ozone1) earth.mod <- earth(O3 ~ ., data=ozone1, degree=2) ev <- evimp(earth.mod, trim=FALSE) plot(ev) print(ev)
Looks like there are no examples yet.