r - 在传递到 `geom_spoke` 之前汇总 2D bin 中的 X、Y、theta 数据

Question

我有包含 x、y 坐标和航向角的数据，我想将它们分成 2D 箱，以便计算每个箱的平均航向并使用 ggplot 的 绘制>geom_spoke.

这是我想要执行的操作示例，其中包含手动创建的垃圾箱:

# data
set.seed(1)
dat <- data.frame(x = runif(100,0,100), y = runif(100,0,100), angle = runif(100, 0, 2*pi))

# manual binning
bins <- rbind(
  #bottom left
  dat %>%
    filter(x < 50 & y < 50) %>%
    summarise(x = 25, y = 25, angle = mean(angle), n = n()),
  #bottom right
  dat %>%
    filter(x > 50 & y < 50) %>%
    summarise(x = 75, y = 25, angle = mean(angle), n = n()),
  #top left
  dat %>%
    filter(x < 50 & y > 50) %>%
    summarise(x = 25, y = 75, angle = mean(angle), n = n()),
  #top right
  dat %>%
    filter(x > 50 & y > 50) %>%
    summarise(x = 75, y = 75, angle = mean(angle), n = n())
)

# plot
ggplot(bins, aes(x, y)) +
  geom_point() +
  coord_equal() +
  scale_x_continuous(limits = c(0,100)) +
  scale_y_continuous(limits = c(0,100)) +
  geom_spoke(aes(angle = angle, radius = n/2), arrow=arrow(length = unit(0.2,"cm")))

我知道如何为每个 bin 创建包含计数数据的 2D bin，例如:

# heatmap of x,y counts
p <- ggplot(dat, aes(x, y)) +
  geom_bin2d(binwidth = c(50, 50)) +
  coord_equal()
#ggplot_build(p)$data[[1]] #access binned data

但我似乎无法找到一种方法来总结其他变量，例如在传递给 geom_spoke 之前每个 bin 的标题。没有第一次装箱，我的情节看起来像这样:

Answer 1

这是一种方法。您需要一次确定每个维度 (x 和 y) 中 bin 的数量/范围，其他所有内容都应包含在代码中:

# adjust range & number of bins here
x.range <- pretty(dat$x, n = 3)
y.range <- pretty(dat$y, n = 3)

> x.range
[1]   0  50 100
> y.range
[1]   0  50 100

根据每一行落入的 x 和 y 间隔自动将每一行分配给一个 bin:

dat <- dat %>%
  rowwise() %>%
  mutate(x.bin = max(which(x > x.range)),
         y.bin = max(which(y > y.range)),
         bin = paste(x.bin, y.bin, sep = "_")) %>%
  ungroup()

> head(dat)
# A tibble: 6 x 6
         x        y    angle x.bin y.bin   bin
     <dbl>    <dbl>    <dbl> <int> <int> <chr>
1 26.55087 65.47239 1.680804     1     2   1_2
2 37.21239 35.31973 1.373789     1     1   1_1
3 57.28534 27.02601 3.247130     2     1   2_1
4 90.82078 99.26841 1.689866     2     2   2_2
5 20.16819 63.34933 1.138314     1     2   1_2
6 89.83897 21.32081 3.258310     2     1   2_1

计算每个 bin 的平均值:

dat <- dat %>%
  group_by(bin) %>%
  mutate(x.mean = mean(x),
         y.mean = mean(y),
         angle.mean = mean(angle),
         n = n()) %>%
  ungroup()

> head(dat)
# A tibble: 6 x 10
         x        y    angle x.bin y.bin   bin   x.mean   y.mean angle.mean     n
     <dbl>    <dbl>    <dbl> <int> <int> <chr>    <dbl>    <dbl>      <dbl> <int>
1 26.55087 65.47239 1.680804     1     2   1_2 26.66662 68.56461   2.672454    29
2 37.21239 35.31973 1.373789     1     1   1_1 33.05887 28.86027   2.173177    23
3 57.28534 27.02601 3.247130     2     1   2_1 74.71214 24.99131   3.071629    23
4 90.82078 99.26841 1.689866     2     2   2_2 77.05622 77.91031   3.007859    25
5 20.16819 63.34933 1.138314     1     2   1_2 26.66662 68.56461   2.672454    29
6 89.83897 21.32081 3.258310     2     1   2_1 74.71214 24.99131   3.071629    23

不对任何 bin 编号/bin 宽度进行硬编码的绘图:

ggplot(dat,
       aes(x, y, fill = bin)) +
  geom_bin2d(binwidth = c(diff(x.range)[1], 
                          diff(y.range)[1])) +
  geom_point(aes(x = x.mean, y = y.mean)) +
  geom_spoke(aes(x = x.mean, y = y.mean, angle = angle.mean, radius = n/2),
             arrow=arrow(length = unit(0.2,"cm"))) +
  coord_equal()

填充调色板的选择、图例标签、绘图标题等其他细节可以随后进行调整。