r - 我如何切割 `geom_curve` 绘制的曲线，以免它们与 `geom_text` 中 `ggplot2` 绘制的标签重叠？

Question

我有一个 df 包含 var 和 link 类型的元素。我想使用 geom_text 将元素 var 绘制为文本标签，使用 geom_curve 将 links 绘制为箭头。问题是，链接 过度绘制了标签，而我希望它们在标签旁边开始和停止。

---

我有什么:

  type   x    y                    label from_x from_y to_x to_y
1 link  NA   NA                     <NA>    608   -229  460 -276
2 link  NA   NA                     <NA>    428   -274  570 -226
3  var 610 -226 accomplishments per week    608   -229  460   NA
4  var 426 -276    hours worked per week    428   -274  570   NA

当我绘制它时，它看起来如下:

ggplot(df) + geom_text(aes(x, y, label = label)) + geom_curve(aes(x = 
from_x,y = from_y,xend = to_x, yend = to_y), curvature = -.3, arrow = arrow(length = unit(0.03, "npc")), color = "red")

---

我期望的是:

我怎样才能做到这一点？

---

这是我的df:

df <- structure(list(type = structure(c(1L, 1L, 2L, 2L), .Label = c("link", 
"var"), class = "factor"), x = c(NA, NA, 610, 426), y = c(NA, 
NA, -226, -276), label = c(NA, NA, "accomplishments per week", 
"hours worked per week"), from_x = c(610, 426, NA, NA), from_y = c(-226, 
-276, NA, NA), to_x = c(426, 610, NA, NA), to_y = c(-276, -226, 
NA, NA)), .Names = c("type", "x", "y", "label", "from_x", "from_y", 
"to_x", "to_y"), row.names = c(NA, -4L), class = "data.frame")

这是我为绘制预期输出所做的手动调整:

df$from_x <- c(608, 428)
df$from_y <- c(-229, -274)
df$to_x <- c(460, 570)

ggplot(df) + geom_text(aes(x, y, label = label)) + geom_curve(aes(x = from_x,y = from_y,xend = to_x, yend = to_y), curvature = -.3, arrow = arrow(length = unit(0.03, "npc")), color = "red")

Answer 1

1. 为 to_x 和 from_y 计算适当的偏移量。
2. 更改 geom 的顺序，使 geom_text 在堆栈中最后呈现(即顶部)

df <-
  df %>%
  mutate(
    to_xoffset = if_else(to_y > -250, to_x - 25, NA_real_),
    to_xoffset = if_else(to_y < -250, to_x + 25, to_xoffset),

    from_yoffset = if_else(from_x < 525, from_y + 2, NA_real_),
    from_yoffset = if_else(from_x > 525, from_y - 2, from_yoffset)
  )

ggplot(df) + 
  geom_curve(aes(x = from_x,y = from_yoffset ,xend = to_xoffset, yend = to_y), curvature = -.3, arrow = arrow(length = unit(0.03, "npc")), color = "red") +
  geom_text(aes(x, y, label = label))