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
|
#' Time Feature Generator
#'
#' `step_time()` creates a *specification* of a recipe
#' step that will convert date-time data into one or more factor or
#' numeric variables.
#'
#' @inheritParams step_pca
#' @inheritParams step_center
#' @param ... One or more selector functions to choose variables
#' for this step. The selected variables should have class
#' `POSIXct` or `POSIXlt`. See [selections()] for more details.
#' @param features A character string that includes at least one
#' of the following values: `am` (is is AM), `hour`, `hour12`, `minute`,
#' `second`, `decimal_day`.
#' @param columns A character string of variables that will be
#' used as inputs. This field is a placeholder and will be
#' populated once [prep()] is used.
#' @param keep_original_cols A logical to keep the original variables in the
#' output. Defaults to `TRUE`.
#' @template step-return
#' @family dummy variable and encoding steps
#' @export
#' @details Unlike some other steps, `step_time()` does *not*
#' remove the original time variables by default. Set `keep_original_cols`
#' to `FALSE` to remove them.
#'
#' `decimal_day` return time of day as a decimal number between 0 and 24. for
#' example `"07:15:00"` would be transformed to `7.25` and `"03:59:59"` would
#' be transformed to `3.999722`. The formula for these calculations are
#' `hour(x) + (second(x) + minute(x) * 60) / 3600`.
#'
#' See [step_date()] if you want to calculate features that are larger than
#' hours.
#'
#' # Tidying
#'
#' When you [`tidy()`][tidy.recipe()] this step, a tibble with columns
#' `terms` (the selectors or variables selected) and `value` (the feature
#' names).
#'
#' @examples
#' library(lubridate)
#'
#' examples <- data.frame(
#' times = ymd_hms("2022-05-06 23:51:07") +
#' hours(1:5) + minutes(1:5) + seconds(1:5)
#' )
#' time_rec <- recipe(~ times, examples) %>%
#' step_time(all_predictors())
#'
#' tidy(time_rec, number = 1)
#'
#' time_rec <- prep(time_rec, training = examples)
#'
#' time_values <- bake(time_rec, new_data = examples)
#' time_values
#'
#' tidy(time_rec, number = 1)
step_time <-
function(recipe,
...,
role = "predictor",
trained = FALSE,
features = c("hour", "minute", "second"),
columns = NULL,
keep_original_cols = TRUE,
skip = FALSE,
id = rand_id("time")) {
feat <-
c(
"am",
"hour",
"hour12",
"minute",
"second",
"decimal_day"
)
if (!is_tune(features)) {
if (!all(features %in% feat)) {
rlang::abort(paste0(
"Possible values of `features` should include: ",
paste0("'", feat, "'", collapse = ", ")
))
}
}
add_step(
recipe,
step_time_new(
terms = enquos(...),
role = role,
trained = trained,
features = features,
columns = columns,
keep_original_cols = keep_original_cols,
skip = skip,
id = id
)
)
}
step_time_new <-
function(terms, role, trained, features, columns, keep_original_cols, skip,
id) {
step(
subclass = "time",
terms = terms,
role = role,
trained = trained,
features = features,
columns = columns,
keep_original_cols = keep_original_cols,
skip = skip,
id = id
)
}
#' @export
prep.step_time <- function(x, training, info = NULL, ...) {
col_names <- recipes_eval_select(x$terms, training, info)
check_type(training[, col_names], types = "datetime")
step_time_new(
terms = x$terms,
role = x$role,
trained = TRUE,
features = x$features,
columns = col_names,
keep_original_cols = get_keep_original_cols(x),
skip = x$skip,
id = x$id
)
}
#' @export
bake.step_time <- function(object, new_data, ...) {
check_new_data(names(object$columns), object, new_data)
for (column in object$columns) {
time_values <- get_time_features(
dt = new_data[[column]],
feats = object$features
)
names(time_values) <- glue::glue("{column}_{names(time_values)}")
new_data <- bind_cols(new_data, time_values)
}
keep_original_cols <- get_keep_original_cols(object)
if (!keep_original_cols) {
new_data <- new_data[, !(colnames(new_data) %in% object$columns), drop = FALSE]
}
new_data
}
get_time_features <- function(dt, feats) {
features <- list(
am = am,
hour = hour,
hour12 = function(x) {
out <- hour(x)
out <- ifelse(out == 0L, 12L, out)
out <- ifelse(out > 12L, out - 12L, out)
out
},
minute = minute,
second = second,
decimal_day = function(x) hour(x) + (second(x) + minute(x) * 60) / 3600
)
purrr::map_dfc(features[feats], ~.x(dt))
}
print.step_time <-
function(x, width = max(20, options()$width - 29), ...) {
title <- "Time features from "
print_step(x$columns, x$terms, x$trained, title, width)
invisible(x)
}
#' @rdname tidy.recipe
#' @export
tidy.step_time <- function(x, ...) {
if (is_trained(x)) {
res <- tidyr::crossing(
terms = unname(x$columns),
value = x$features
)
} else {
term_names <- sel2char(x$terms)
res <- tidyr::crossing(
terms = term_names,
value = x$features
)
}
tibble::add_column(res, id = x$id)
}
|
