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
|
% Generated by roxygen2: do not edit by hand
% Please edit documentation in R/varsel.R
\name{varsel}
\alias{varsel}
\alias{varsel.default}
\alias{varsel.refmodel}
\title{Variable selection for generalized linear models}
\usage{
varsel(object, ...)
\method{varsel}{default}(object, ...)
\method{varsel}{refmodel}(
object,
d_test = NULL,
method = NULL,
ndraws = NULL,
nclusters = NULL,
ndraws_pred = NULL,
nclusters_pred = NULL,
cv_search = TRUE,
nterms_max = NULL,
intercept = TRUE,
verbose = TRUE,
lambda_min_ratio = 1e-05,
nlambda = 150,
thresh = 1e-06,
regul = 1e-04,
penalty = NULL,
search_terms = NULL,
...
)
}
\arguments{
\item{object}{Either a \code{refmodel}-type object created by
\link[=get_refmodel]{get_refmodel}, a \link[=init_refmodel]{init_refmodel},
an object which can be converted to a reference model using
\link[=get_refmodel]{get_refmodel} or a \code{vsel} object resulting from
\code{varsel} or \code{cv_varsel}.}
\item{...}{Additional arguments to be passed to the
\code{get_refmodel}-function.}
\item{d_test}{A test dataset, which is used to evaluate model performance. If
not provided, training data is used. Currently this argument is for
internal use only.}
\item{method}{The method used in the variable selection. Possible options are
\code{'L1'} for L1-search and \code{'forward'} for forward selection.
Default is 'forward' if the number of variables in the full data is at most
20,' and \code{'L1'} otherwise.}
\item{ndraws}{Number of posterior draws used in the variable selection.
Cannot be larger than the number of draws in the reference model. Ignored
if nclusters is set.}
\item{nclusters}{Number of clusters to use in the clustered projection.
Overrides the \code{ndraws} argument. Defaults to 1.}
\item{ndraws_pred}{Number of projected draws used for prediction (after
selection). Ignored if nclusters_pred is given. Note that setting less
draws or clusters than posterior draws in the reference model may result in
slightly inaccurate projection performance, although increasing this
argument linearly affects the computation time.}
\item{nclusters_pred}{Number of clusters used for prediction (after
selection). Default is 5.}
\item{cv_search}{If TRUE, then the projected coefficients after L1-selection
are computed without any penalization (or using only the regularization
determined by \code{regul}). If FALSE, then the coefficients are the
solution from the' L1-penalized projection. This option is relevant only if
\code{method}='L1'. Default is TRUE for genuine reference models and FALSE
if \code{object} is datafit (see \link[=init_refmodel]{init_refmodel}).}
\item{nterms_max}{Maximum number of varibles until which the selection is
continued. Defaults to min(20, D, floor(0.4*n)) where n is the number of
observations and D the number of variables.}
\item{intercept}{Whether to use intercept in the submodels. Defaults to TRUE.}
\item{verbose}{If TRUE, may print out some information during the selection.
Defaults to FALSE.}
\item{lambda_min_ratio}{Ratio between the smallest and largest lambda in the
L1-penalized search. This parameter essentially determines how long the
search is carried out, i.e., how large submodels are explored. No need to
change the default value unless the program gives a warning about this.}
\item{nlambda}{Number of values in the lambda grid for L1-penalized search.
No need to change unless the program gives a warning about this.}
\item{thresh}{Convergence threshold when computing L1-path. Usually no need
to change this.}
\item{regul}{Amount of regularization in the projection. Usually there is no
need for regularization, but sometimes for some models the projection can
be ill-behaved and we need to add some regularization to avoid numerical
problems.}
\item{penalty}{Vector determining the relative penalties or costs for the
variables. Zero means that those variables have no cost and will therefore
be selected first, whereas Inf means those variables will never be
selected. Currently works only if method == 'L1'. By default 1 for each
variable.}
\item{search_terms}{A custom list of terms to evaluate for variable
selection. By default considers all the terms in the reference model's
formula.}
}
\value{
An object of type \code{vsel} that contains information about the
feature selection. The fields are not meant to be accessed directly by
the user but instead via the helper functions (see the vignettes or type
?projpred to see the main functions in the package.)
}
\description{
Perform the projection predictive variable selection for generalized linear
models, generalized linear and additive multilevel models using generic
reference models.
}
\examples{
\donttest{
if (requireNamespace('rstanarm', quietly=TRUE)) {
### Usage with stanreg objects
n <- 30
d <- 5
x <- matrix(rnorm(n*d), nrow=n)
y <- x[,1] + 0.5*rnorm(n)
data <- data.frame(x,y)
fit <- rstanarm::stan_glm(y ~ X1 + X2 + X3 + X4 + X5, gaussian(), data=data,
chains=2, iter=500)
vs <- varsel(fit)
plot(vs)
}
}
}
|
