我正在寻找一种在不使用coord_cartesian(clip = "off")
的情况下向ggplots添加次要刻度线的方法。或一种可复制的方法将剪裁应用于x轴,而不是y轴,反之亦然。
到目前为止,我一直在使用annotation_ticks()
函数和此出色答案GeomTicks
中定义的here geom(进行了一些小的修改,使其可以与ggplot2 v3.3.0
一起使用)。不幸的是,要使刻度线注释显示在图的外部,必须使用coord_cartesian(clip = "off")
,这意味着位于图区域之外的其他任何内容也将暴露(请参阅下面的reprex)。
[或者,也许有一种方法可以利用ggplot2 v3.3.0
的任何新功能来绘制次要刻度,而不是作为注释,而是作为轴/图的实际部分,以便可以将它们绘制在绘图之外区域。
我不是软件开发人员,但也许可以使用名为register_theme_elements
的axis.minor.ticks
定义一个新的主题元素,其行为类似于axis.ticks
,但从panel_params$y$break_positions_minor
而不是panel_params$y$break_positions
获取次要刻度线的适当位置。或以某种方式使用新的guide_x()
S3功能。
任何帮助将不胜感激!
[annotation_ticks()
功能(合并了this fix以进行构面问题):
annotation_ticks <- function(sides = "b",
scale = "identity",
scaled = TRUE,
ticklength = unit(0.1, "cm"),
colour = "black",
size = 0.5,
linetype = 1,
alpha = 1,
color = NULL,
ticks_per_base = NULL,
data = data.frame(x = NA),
...) {
if (!is.null(color)) {
colour <- color
}
# check for invalid side
if (grepl("[^btlr]", sides)) {
stop(gsub("[btlr]", "", sides), " is not a valid side: b,t,l,r are valid")
}
# split sides to character vector
sides <- strsplit(sides, "")[[1]]
if (length(sides) != length(scale)) {
if (length(scale) == 1) {
scale <- rep(scale, length(sides))
} else {
stop("Number of scales does not match the number of sides")
}
}
base <- sapply(scale, function(x) switch(x, "identity" = 10, "log10" = 10, "log" = exp(1)), USE.NAMES = FALSE)
if (missing(ticks_per_base)) {
ticks_per_base <- base - 1
} else {
if ((length(sides) != length(ticks_per_base))) {
if (length(ticks_per_base) == 1) {
ticks_per_base <- rep(ticks_per_base, length(sides))
} else {
stop("Number of ticks_per_base does not match the number of sides")
}
}
}
delog <- scale %in% "identity"
layer(
data = data,
mapping = NULL,
stat = StatIdentity,
geom = GeomTicks,
position = PositionIdentity,
show.legend = FALSE,
inherit.aes = FALSE,
params = list(
base = base,
sides = sides,
scaled = scaled,
ticklength = ticklength,
colour = colour,
size = size,
linetype = linetype,
alpha = alpha,
ticks_per_base = ticks_per_base,
delog = delog,
...
)
)
}
ggproto对象(现在可用于ggplot2 v3.3.0
:
GeomTicks <- ggproto(
"GeomTicks", Geom,
extra_params = "",
handle_na = function(data, params) {
data
},
draw_panel = function(data,
panel_scales,
coord,
base = c(10, 10),
sides = c("b", "l"),
scaled = TRUE,
ticklength = unit(0.1, "cm"),
ticks_per_base = base - 1,
delog = c(x = TRUE, y = TRUE)) {
ticks <- list()
for (s in 1:length(sides)) {
if (grepl("[b|t]", sides[s])) {
# for ggplot2 < 3.3.0 use: xticks <- panel_params$x.minor
if (utils::packageVersion("ggplot2") >= "3.2.1.9000") {
x_minor_breaks <- panel_scales$x$break_positions_minor()
x_major_breaks <- panel_scales$x$break_positions()
} else {
x_minor_breaks <- panel_scales$x.minor
x_major_breaks <- panel_scales$x.major
}
xticks <- setdiff(x_minor_breaks, x_major_breaks)
# Make the grobs
if (grepl("b", sides[s])) {
ticks$x_b <- with(
data,
segmentsGrob(
x0 = unit(xticks, "npc"),
x1 = unit(xticks, "npc"),
y0 = unit(0, "npc"),
y1 = ticklength,
gp = gpar(
col = alpha(colour, alpha),
lty = linetype,
lwd = size * .pt
)
)
)
}
if (grepl("t", sides[s])) {
ticks$x_t <- with(
data,
segmentsGrob(
x0 = unit(xticks, "npc"),
x1 = unit(xticks, "npc"),
y0 = unit(1, "npc"),
y1 = unit(1, "npc") - ticklength,
gp = gpar(
col = alpha(colour, alpha),
lty = linetype,
lwd = size * .pt
)
)
)
}
}
if (grepl("[l|r]", sides[s])) {
# for ggplot2 < 3.3.0 use: yticks <- panel_params$y.minor
if (utils::packageVersion("ggplot2") >= "3.2.1.9000") {
y_minor_breaks <- panel_scales$y$break_positions_minor()
y_major_breaks <- panel_scales$y$break_positions()
} else {
y_minor_breaks <- panel_scales$y.minor
y_major_breaks <- panel_scales$y.major
}
yticks <- setdiff(y_minor_breaks, y_major_breaks)
# Make the grobs
if (grepl("l", sides[s])) {
ticks$y_l <- with(
data,
segmentsGrob(
y0 = unit(yticks, "npc"),
y1 = unit(yticks, "npc"),
x0 = unit(0, "npc"),
x1 = ticklength,
gp = gpar(
col = alpha(colour, alpha),
lty = linetype, lwd = size * .pt
)
)
)
}
if (grepl("r", sides[s])) {
ticks$y_r <- with(
data,
segmentsGrob(
y0 = unit(yticks, "npc"),
y1 = unit(yticks, "npc"),
x0 = unit(1, "npc"),
x1 = unit(1, "npc") - ticklength,
gp = gpar(
col = alpha(colour, alpha),
lty = linetype,
lwd = size * .pt
)
)
)
}
}
}
gTree(children = do.call("gList", ticks))
},
default_aes = aes(colour = "black", size = 0.5, linetype = 1, alpha = 1)
)
coord_cartesian(clip = "on")
的图具有非常粗的线宽的列看起来不错,但是看不到刻度注释。
library(ggplot2)
library(grid)
ggplot(mpg, aes(x = class, y = displ, fill = class)) +
stat_summary(fun = mean, geom = "col", colour = "black", size = 1) +
theme_classic(base_size = 8) +
scale_y_continuous(limits = c(0, 8), expand = c(0, 0)) +
annotation_ticks(sides = "l", ticklength = -1 * unit(0.2, "cm")) +
coord_cartesian(clip = "on")
ggsave("clip_on.png", device = "png", width = 4, height = 3)
coord_cartesian(clip = "off")
的图刻度线注释可见,但线条宽度很粗的列显示在绘图区域之外。
ggplot(mpg, aes(x = class, y = displ, fill = class)) +
stat_summary(fun = mean, geom = "col", colour = "black", size = 1) +
theme_classic(base_size = 8) +
scale_y_continuous(limits = c(0, 8), expand = c(0, 0)) +
annotation_ticks(sides = "l", ticklength = -1 * unit(0.2, "cm")) +
coord_cartesian(clip = "off")
ggsave("clip_off.png", device = "png", width = 4, height = 3)
这段代码对我来说似乎太熟悉了,所以我想考虑一下。
是的,使用ggplot v3.3.0指南已经变得可扩展,尽管我怀疑它们会长期保持这种状态,因为我听说通过葡萄藤他们也希望将指南切换到ggproto系统。
没有太多风吹草动的最便宜的方法来完成您所要求的,就是调整向导的向导培训部分。由于这是一个S3方法,因此我们需要一个新的指南类来编写自定义方法:
library(ggplot2)
library(rlang)
#> Warning: package 'rlang' was built under R version 3.6.3
library(glue)
guide_axis_minor <- function(
title = waiver(), check.overlap = FALSE, angle = NULL,
n.dodge = 1, order = 0, position = waiver()
) {
structure(list(title = title, check.overlap = check.overlap,
angle = angle, n.dodge = n.dodge, order = order, position = position,
available_aes = c("x", "y"), name = "axis"),
class = c("guide", "axis_minor", "axis"))
}
您会注意到,除了一个额外的类以外,以上函数与guide_axis()
相同。在这里,类的顺序很重要,因为我们在继承axis
类的子类,因此我们可以变得很懒,只使用所有已经存在的方法即可。
这使我们接受培训,这确实是唯一需要调整的地方。我已经评论了相关内容。大多数功能仍与guide_train.axis
内部功能相同。简要地说,我们将小休会视为带有空标签的大休会。
guide_train.axis_minor <- function(guide, scale, aesthetic = NULL) {
aesthetic <- aesthetic %||% scale$aesthetics[1]
# Seperately define major and minor breaks
major_breaks <- scale$get_breaks()
minor_breaks <- scale$get_breaks_minor()
# We set the actual breaks to be both major and minor
breaks <- union(major_breaks, minor_breaks)
# We keep track of what breaks were the major breaks
is_major <- breaks %in% major_breaks
empty_ticks <- ggplot2:::new_data_frame(
list(aesthetic = numeric(), .value = numeric(0), .label = character())
)
if (length(intersect(scale$aesthetics, guide$available_aes)) == 0) {
warn(glue("axis guide needs appropriate scales: ",
glue_collapse(guide$available_aes, ", ", last = " or ")))
guide$key <- empty_ticks
} else if (length(breaks) == 0) {
guide$key <- empty_ticks
} else {
mapped_breaks <- if (scale$is_discrete()) {
scale$map(breaks)
} else {
breaks
}
ticks <- ggplot2:::new_data_frame(setNames(list(mapped_breaks),
aesthetic))
ticks$.value <- breaks
ticks$.label <- scale$get_labels(breaks)
# Now this is the bit where we set minor breaks to have empty labls
ticks$.label[!is_major] <- ""
guide$key <- ticks[is.finite(ticks[[aesthetic]]), ]
}
guide$name <- paste0(guide$name, "_", aesthetic)
guide$hash <- digest::digest(list(guide$title, guide$key$.value,
guide$key$.label, guide$name))
guide
}
现在您可以按名称从任何连续位置比例调用该指南:
ggplot(mpg, aes(x = class, y = displ, fill = class)) +
stat_summary(fun = mean, geom = "col") +
scale_y_continuous(limits = c(0, 8),
guide = "axis_minor")
由reprex package(v0.3.0)在2020-04-07创建