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
32
33
|
#' @export
makeRLearner.regr.pcr = function() {
makeRLearnerRegr(
cl = "regr.pcr",
package = "pls",
par.set = makeParamSet(
makeIntegerLearnerParam(id = "ncomp", lower = 1L),
makeDiscreteLearnerParam(id = "method", default = "cppls",
values = c("kernelpls", "widekernelpls", "simpls", "oscorespls", "cppls", "svdpc")),
makeLogicalLearnerParam(id = "scale", default = FALSE),
makeLogicalLearnerParam(id = "model", default = TRUE, tunable = FALSE),
makeLogicalLearnerParam(id = "x", default = FALSE, tunable = FALSE),
makeLogicalLearnerParam(id = "y", default = FALSE, tunable = FALSE)
),
par.vals = list(model = FALSE),
properties = c("numerics", "factors"),
name = "Principal Component Regression",
short.name = "pcr",
callees = "pcr"
)
}
#' @export
trainLearner.regr.pcr = function(.learner, .task, .subset, .weights = NULL, ...) {
f = getTaskFormula(.task)
pls::pcr(f, data = getTaskData(.task, .subset), ...)
}
#' @export
predictLearner.regr.pcr = function(.learner, .model, .newdata, ...) {
p = predict(.model$learner.model, newdata = .newdata)
p[, 1L, dim(p)[3L]]
}
|
