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
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
|
plot_type_int <- function(model,
mdrt.values,
ci.lvl,
pred.type,
facets,
show.data,
jitter,
geom.colors,
axis.title,
title,
legend.title,
axis.lim,
case,
show.legend,
dot.size,
line.size,
...) {
# interaction terms are separated with ":"
int.terms <- insight::find_interactions(model, component = "conditional", flatten = TRUE)
# stop if no interaction found
if (is.null(int.terms))
stop("No interaction term found in model.", call. = F)
# get interaction terms and model frame
ia.terms <- purrr::map(int.terms, ~ sjmisc::trim(unlist(strsplit(.x, "[\\*:]"))))
mf <- insight::get_data(model, verbose = FALSE)
pl <- list()
# intertate interaction terms
for (i in seq_along(ia.terms)) {
ia <- ia.terms[[i]]
find.fac <- purrr::map_lgl(ia, ~ is_categorical(mf[[.x]]))
# find all non-categorical variables, except first
# term, which is considered as being along the x-axis
check_cont <- ia[-1][!find.fac[2:length(find.fac)]]
# if we have just categorical as interaction terms,
# we plot all category values
if (!sjmisc::is_empty(check_cont)) {
# get data from continuous interaction terms. we
# need this to compute the specific values that
# should be used as group characteristic for the plot
cont_terms <- dplyr::select(mf, !! check_cont)
# for quartiles used as moderator values, make sure
# that the variable's range is large enough to compute
# quartiles
mdrt.val <- mv_check(mdrt.values = mdrt.values, cont_terms)
# prepare terms for ggpredict()-call. terms is a character-vector
# with term name and values to plot in square brackets
terms <- purrr::map_chr(check_cont, function(x) {
if (mdrt.val == "minmax") {
ct.min <- min(cont_terms[[x]], na.rm = TRUE)
ct.max <- max(cont_terms[[x]], na.rm = TRUE)
if (sjmisc::is_float(ct.min) || sjmisc::is_float(ct.max))
sprintf("%s [%.2f,%.2f]", x, ct.min, ct.max)
else
sprintf("%s [%i,%i]", x, ct.min, ct.max)
} else if (mdrt.val == "meansd") {
mw <- mean(cont_terms[[x]], na.rm = TRUE)
sabw <- stats::sd(cont_terms[[x]], na.rm = TRUE)
sprintf("%s [%.2f,%.2f,%.2f]", x, mw, mw - sabw, mw + sabw)
} else if (mdrt.val == "zeromax") {
ct.max <- max(cont_terms[[x]], na.rm = TRUE)
if (sjmisc::is_float(ct.max))
sprintf("%s [0,%.2f]", x, ct.max)
else
sprintf("%s [0,%i]", x, ct.max)
} else if (mdrt.val == "quart") {
qu <- as.vector(stats::quantile(cont_terms[[x]], na.rm = T))
sprintf("%s [%.2f,%.2f,%.2f]", x, qu[3], qu[2], qu[4])
} else {
x
}
})
ia[match(check_cont, ia)] <- terms
}
# compute marginal effects for interaction terms
pred.type <- switch(pred.type,
fe = "fixed",
re = "random",
pred.type
)
dat <- ggeffects::ggpredict(
model = model,
terms = ia,
ci_level = ci.lvl,
type = pred.type,
full.data = FALSE,
...
)
# evaluate dots-arguments
alpha <- .15
dodge <- .1
dot.alpha <- .5
log.y <- FALSE
add.args <- lapply(match.call(expand.dots = F)$`...`, function(x) x)
if ("alpha" %in% names(add.args)) alpha <- eval(add.args[["alpha"]])
if ("dodge" %in% names(add.args)) dodge <- eval(add.args[["dodge"]])
if ("dot.alpha" %in% names(add.args)) dot.alpha <- eval(add.args[["dot.alpha"]])
if ("log.y" %in% names(add.args)) log.y <- eval(add.args[["log.y"]])
# select color palette
if (is.null(geom.colors) || geom.colors[1] != "bw")
geom.colors <- col_check2(geom.colors, dplyr::n_distinct(dat$group))
# save plot of marginal effects for interaction terms
p <- graphics::plot(
dat,
show_ci = !is.na(ci.lvl),
facets = facets,
show_data = show.data,
colors = geom.colors,
jitter = jitter,
use_theme = FALSE,
case = case,
show_legend = show.legend,
dot_alpha = dot.alpha,
alpha = alpha,
dodge = dodge,
log_y = log.y,
dot_size = dot.size,
line_size = line.size
)
# set axis and plot titles
if (!is.null(axis.title)) {
if (length(axis.title) > 1) {
p <- p + labs(x = axis.title[1], y = axis.title[2])
} else {
p <- p + labs(y = axis.title)
}
}
# set axis and plot titles
if (!is.null(title))
p <- p + ggtitle(title)
# set axis and plot titles
if (!is.null(legend.title))
p <- p + labs(colour = legend.title)
# set axis limits
if (!is.null(axis.lim)) {
if (is.list(axis.lim))
p <- p + xlim(axis.lim[[1]]) + ylim(axis.lim[[2]])
else
p <- p + ylim(axis.lim)
}
# add plot result to final return value
if (length(ia.terms) == 1)
pl <- p
else
pl[[length(pl) + 1]] <- p
}
pl
}
#' @importFrom stats na.omit
is_categorical <- function(x) {
is.factor(x) || (length(unique(stats::na.omit(x))) < 3)
}
#' @importFrom stats quantile
#' @importFrom purrr map_dbl
mv_check <- function(mdrt.values, x) {
# for quartiles used as moderator values, make sure
# that the variable's range is large enough to compute
# quartiles
if (mdrt.values == "quart") {
if (!is.data.frame(x)) x <- as.data.frame(x)
mvc <- purrr::map_dbl(x, ~ length(unique(as.vector(stats::quantile(.x, na.rm = T)))))
if (any(mvc < 3)) {
# tell user that quart won't work
message("Could not compute quartiles, too small range of moderator variable. Defaulting `mdrt.values` to `minmax`.")
mdrt.values <- "minmax"
}
}
mdrt.values
}
|
