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
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
|
% Generated by roxygen2: do not edit by hand
% Please edit documentation in R/gafs.R, R/safs.R
\name{gafsControl}
\alias{gafsControl}
\alias{safsControl}
\title{Control parameters for GA and SA feature selection}
\usage{
gafsControl(
functions = NULL,
method = "repeatedcv",
metric = NULL,
maximize = NULL,
number = ifelse(grepl("cv", method), 10, 25),
repeats = ifelse(grepl("cv", method), 1, 5),
verbose = FALSE,
returnResamp = "final",
p = 0.75,
index = NULL,
indexOut = NULL,
seeds = NULL,
holdout = 0,
genParallel = FALSE,
allowParallel = TRUE
)
safsControl(
functions = NULL,
method = "repeatedcv",
metric = NULL,
maximize = NULL,
number = ifelse(grepl("cv", method), 10, 25),
repeats = ifelse(grepl("cv", method), 1, 5),
verbose = FALSE,
returnResamp = "final",
p = 0.75,
index = NULL,
indexOut = NULL,
seeds = NULL,
holdout = 0,
improve = Inf,
allowParallel = TRUE
)
}
\arguments{
\item{functions}{a list of functions for model fitting, prediction etc (see
Details below)}
\item{method}{The resampling method: \code{boot}, \code{boot632}, \code{cv},
\code{repeatedcv}, \code{LOOCV}, \code{LGOCV} (for repeated training/test
splits)}
\item{metric}{a two-element string that specifies what summary metric will
be used to select the optimal number of iterations from the external fitness
value and which metric should guide subset selection. If specified, this
vector should have names \code{"internal"} and \code{"external"}. See
\code{\link{gafs}} and/or \code{\link{safs}} for explanations of the
difference.}
\item{maximize}{a two-element logical: should the metrics be maximized or
minimized? Like the \code{metric} argument, this this vector should have
names \code{"internal"} and \code{"external"}.}
\item{number}{Either the number of folds or number of resampling iterations}
\item{repeats}{For repeated k-fold cross-validation only: the number of
complete sets of folds to compute}
\item{verbose}{a logical for printing results}
\item{returnResamp}{A character string indicating how much of the resampled
summary metrics should be saved. Values can be ``all'' or ``none''}
\item{p}{For leave-group out cross-validation: the training percentage}
\item{index}{a list with elements for each resampling iteration. Each list
element is the sample rows used for training at that iteration.}
\item{indexOut}{a list (the same length as \code{index}) that dictates which
sample are held-out for each resample. If \code{NULL}, then the unique set
of samples not contained in \code{index} is used.}
\item{seeds}{a vector or integers that can be used to set the seed during
each search. The number of seeds must be equal to the number of resamples
plus one.}
\item{holdout}{the proportion of data in [0, 1) to be held-back from
\code{x} and \code{y} to calculate the internal fitness values}
\item{genParallel}{if a parallel backend is loaded and available, should
\code{\link{gafs}} use it tp parallelize the fitness calculations within a
generation within a resample?}
\item{allowParallel}{if a parallel backend is loaded and available, should
the function use it?}
\item{improve}{the number of iterations without improvement before
\code{\link{safs}} reverts back to the previous optimal subset}
}
\value{
An echo of the parameters specified
}
\description{
Control the computational nuances of the \code{\link{gafs}} and
\code{\link{safs}} functions
Many of these options are the same as those described for
\code{\link[caret]{trainControl}}. More extensive documentation and examples
can be found on the \pkg{caret} website at
\url{http://topepo.github.io/caret/feature-selection-using-genetic-algorithms.html#syntax} and
\url{http://topepo.github.io/caret/feature-selection-using-simulated-annealing.html#syntax}.
The \code{functions} component contains the information about how the model
should be fit and summarized. It also contains the elements needed for the
GA and SA modules (e.g. cross-over, etc).
The elements of \code{functions} that are the same for GAs and SAs are:
\itemize{
\item \code{fit}, with arguments \code{x}, \code{y}, \code{lev},
\code{last}, and \code{...}, is used to fit the classification or regression
model
\item \code{pred}, with arguments \code{object} and \code{x}, predicts
new samples
\item \code{fitness_intern}, with arguments \code{object},
\code{x}, \code{y}, \code{maximize}, and \code{p}, summarizes performance
for the internal estimates of fitness
\item \code{fitness_extern}, with
arguments \code{data}, \code{lev}, and \code{model}, summarizes performance
using the externally held-out samples
\item \code{selectIter}, with
arguments \code{x}, \code{metric}, and \code{maximize}, determines the best
search iteration for feature selection.
}
The elements of \code{functions} specific to genetic algorithms are:
\itemize{
\item \code{initial}, with arguments \code{vars}, \code{popSize}
and \code{...}, creates an initial population.
\item \code{selection}, with
arguments \code{population}, \code{fitness}, \code{r}, \code{q}, and
\code{...}, conducts selection of individuals.
\item \code{crossover}, with
arguments \code{population}, \code{fitness}, \code{parents} and \code{...},
control genetic reproduction.
\item \code{mutation}, with arguments
\code{population}, \code{parent} and \code{...}, adds mutations.
}
The elements of \code{functions} specific to simulated annealing are:
\itemize{
\item \code{initial}, with arguments \code{vars}, \code{prob}, and
\code{...}, creates the initial subset.
\item \code{perturb}, with
arguments \code{x}, \code{vars}, and \code{number}, makes incremental
changes to the subsets.
\item \code{prob}, with arguments \code{old},
\code{new}, and \code{iteration}, computes the acceptance probabilities
}
The pages \url{http://topepo.github.io/caret/feature-selection-using-genetic-algorithms.html} and
\url{http://topepo.github.io/caret/feature-selection-using-simulated-annealing.html} have more details about each of
these functions.
\code{holdout} can be used to hold out samples for computing the internal
fitness value. Note that this is independent of the external resampling
step. Suppose 10-fold CV is being used. Within a resampling iteration,
\code{holdout} can be used to sample an additional proportion of the 90\%
resampled data to use for estimating fitness. This may not be a good idea
unless you have a very large training set and want to avoid an internal
resampling procedure to estimate fitness.
The search algorithms can be parallelized in several places:
\enumerate{
\item each externally resampled GA or SA can be run independently
(controlled by the \code{allowParallel} options)
\item within a GA, the
fitness calculations at a particular generation can be run in parallel over
the current set of individuals (see the \code{genParallel})
\item if inner resampling is used, these can be run in parallel (controls depend on the
function used. See, for example, \code{\link[caret]{trainControl}})
\item any parallelization of the individual model fits. This is also specific to the modeling function.
}
It is probably best to pick one of these areas for parallelization and the
first is likely to produces the largest decrease in run-time since it is the
least likely to incur multiple re-starting of the worker processes. Keep in
mind that if multiple levels of parallelization occur, this can effect the
number of workers and the amount of memory required exponentially.
}
\references{
\url{http://topepo.github.io/caret/feature-selection-using-genetic-algorithms.html},
\url{http://topepo.github.io/caret/feature-selection-using-simulated-annealing.html}
}
\seealso{
\code{\link{safs}}, \code{\link{safs}}, , \code{\link{caretGA}},
\code{\link{rfGA}}, \code{\link{treebagGA}}, \code{\link{caretSA}},
\code{\link{rfSA}}, \code{\link{treebagSA}}
}
\author{
Max Kuhn
}
\keyword{utilities}
|
