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
|
% Generated by roxygen2: do not edit by hand
% Please edit documentation in R/normalize.R, R/unnormalize.R
\name{normalize}
\alias{normalize}
\alias{normalize.numeric}
\alias{normalize.data.frame}
\alias{unnormalize}
\alias{unnormalize.numeric}
\alias{unnormalize.data.frame}
\title{Normalize numeric variable to 0-1 range}
\usage{
normalize(x, ...)
\method{normalize}{numeric}(x, include_bounds = TRUE, verbose = TRUE, ...)
\method{normalize}{data.frame}(
x,
select = NULL,
exclude = NULL,
include_bounds = TRUE,
ignore_case = FALSE,
regex = FALSE,
verbose = TRUE,
...
)
unnormalize(x, ...)
\method{unnormalize}{numeric}(x, verbose = TRUE, ...)
\method{unnormalize}{data.frame}(
x,
select = NULL,
exclude = NULL,
ignore_case = FALSE,
regex = FALSE,
verbose = TRUE,
...
)
}
\arguments{
\item{x}{A numeric vector, (grouped) data frame, or matrix. See 'Details'.}
\item{...}{Arguments passed to or from other methods.}
\item{include_bounds}{Numeric or logical. Using this can be useful in case of
beta-regression, where the response variable is not allowed to include
zeros and ones. If \code{TRUE}, the input is normalized to a range that includes
zero and one. If \code{FALSE}, the return value is compressed, using
Smithson and Verkuilen's (2006) formula \code{(x * (n - 1) + 0.5) / n}, to avoid
zeros and ones in the normalized variables. Else, if numeric (e.g., \code{0.001}),
\code{include_bounds} defines the "distance" to the lower and upper bound, i.e.
the normalized vectors are rescaled to a range from \code{0 + include_bounds} to
\code{1 - include_bounds}.}
\item{verbose}{Toggle warnings and messages on or off.}
\item{select}{Variables that will be included when performing the required
tasks. Can be either
\itemize{
\item a variable specified as a literal variable name (e.g., \code{column_name}),
\item a string with the variable name (e.g., \code{"column_name"}), or a character
vector of variable names (e.g., \code{c("col1", "col2", "col3")}),
\item a formula with variable names (e.g., \code{~column_1 + column_2}),
\item a vector of positive integers, giving the positions counting from the left
(e.g. \code{1} or \code{c(1, 3, 5)}),
\item a vector of negative integers, giving the positions counting from the
right (e.g., \code{-1} or \code{-1:-3}),
\item one of the following select-helpers: \code{starts_with()}, \code{ends_with()},
\code{contains()}, a range using \code{:} or \code{regex("")}. \code{starts_with()},
\code{ends_with()}, and \code{contains()} accept several patterns, e.g
\code{starts_with("Sep", "Petal")}.
\item or a function testing for logical conditions, e.g. \code{is.numeric()} (or
\code{is.numeric}), or any user-defined function that selects the variables
for which the function returns \code{TRUE} (like: \code{foo <- function(x) mean(x) > 3}),
\item ranges specified via literal variable names, select-helpers (except
\code{regex()}) and (user-defined) functions can be negated, i.e. return
non-matching elements, when prefixed with a \code{-}, e.g. \code{-ends_with("")},
\code{-is.numeric} or \code{-Sepal.Width:Petal.Length}. \strong{Note:} Negation means
that matches are \emph{excluded}, and thus, the \code{exclude} argument can be
used alternatively. For instance, \code{select=-ends_with("Length")} (with
\code{-}) is equivalent to \code{exclude=ends_with("Length")} (no \code{-}). In case
negation should not work as expected, use the \code{exclude} argument instead.
}
If \code{NULL}, selects all columns. Patterns that found no matches are silently
ignored, e.g. \code{find_columns(iris, select = c("Species", "Test"))} will just
return \code{"Species"}.}
\item{exclude}{See \code{select}, however, column names matched by the pattern
from \code{exclude} will be excluded instead of selected. If \code{NULL} (the default),
excludes no columns.}
\item{ignore_case}{Logical, if \code{TRUE} and when one of the select-helpers or
a regular expression is used in \code{select}, ignores lower/upper case in the
search pattern when matching against variable names.}
\item{regex}{Logical, if \code{TRUE}, the search pattern from \code{select} will be
treated as regular expression. When \code{regex = TRUE}, select \emph{must} be a
character string (or a variable containing a character string) and is not
allowed to be one of the supported select-helpers or a character vector
of length > 1. \code{regex = TRUE} is comparable to using one of the two
select-helpers, \code{select = contains("")} or \code{select = regex("")}, however,
since the select-helpers may not work when called from inside other
functions (see 'Details'), this argument may be used as workaround.}
}
\value{
A normalized object.
}
\description{
Performs a normalization of data, i.e., it scales variables in the range
0 - 1. This is a special case of \code{\link[=rescale]{rescale()}}. \code{unnormalize()} is the
counterpart, but only works for variables that have been normalized with
\code{normalize()}.
}
\details{
\itemize{
\item If \code{x} is a matrix, normalization is performed across all values (not
column- or row-wise). For column-wise normalization, convert the matrix to a
data.frame.
\item If \code{x} is a grouped data frame (\code{grouped_df}), normalization is performed
separately for each group.
}
}
\section{Selection of variables - the \code{select} argument}{
For most functions that have a \code{select} argument (including this function),
the complete input data frame is returned, even when \code{select} only selects
a range of variables. That is, the function is only applied to those variables
that have a match in \code{select}, while all other variables remain unchanged.
In other words: for this function, \code{select} will not omit any non-included
variables, so that the returned data frame will include all variables
from the input data frame.
}
\examples{
normalize(c(0, 1, 5, -5, -2))
normalize(c(0, 1, 5, -5, -2), include_bounds = FALSE)
# use a value defining the bounds
normalize(c(0, 1, 5, -5, -2), include_bounds = .001)
head(normalize(trees))
}
\references{
Smithson M, Verkuilen J (2006). A Better Lemon Squeezer? Maximum-Likelihood
Regression with Beta-Distributed Dependent Variables. Psychological Methods,
11(1), 54–71.
}
\seealso{
See \code{\link[=makepredictcall.dw_transformer]{makepredictcall.dw_transformer()}} for use in model formulas.
Other transform utilities:
\code{\link{ranktransform}()},
\code{\link{rescale}()},
\code{\link{reverse}()},
\code{\link{standardize}()}
}
\concept{transform utilities}
|
