1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
|
#' @export
makeRLearner.regr.mars = function() {
makeRLearnerRegr(
cl = "regr.mars",
package = "mda",
par.set = makeParamSet(
makeIntegerLearnerParam(id = "degree", default = 1L, lower = 1L),
makeIntegerLearnerParam(id = "nk", lower = 1L),
makeNumericLearnerParam(id = "penalty", default = 2, lower = 0),
makeNumericLearnerParam(id = "thresh", default = 0.001, lower = 0),
makeLogicalLearnerParam(id = "prune", default = TRUE),
makeLogicalLearnerParam(id = "trace.mars", default = FALSE, tunable = FALSE),
makeLogicalLearnerParam(id = "forward.step", default = TRUE)
),
properties = "numerics",
name = "Multivariate Adaptive Regression Splines",
short.name = "mars",
callees = "mars"
)
}
#' @export
trainLearner.regr.mars = function(.learner, .task, .subset, .weights = NULL, ...) {
d = getTaskData(.task, .subset, target.extra = TRUE)
mda::mars(x = as.matrix(d$data), y = d$target, ...)
}
#' @export
predictLearner.regr.mars = function(.learner, .model, .newdata, ...) {
predict(.model$learner.model, newdata = .newdata)[, 1L]
}
|
