grouped_scatter

数据与公共样式

iris 已收进 Tessera Toy。Petal length 与 Petal width 能让三个物种形成清楚但并非完全无重叠的点群，比 Orange 的重复测量结构更适合演示分组散点。

library(ggplot2)
library(ggpubr)
library(biopalette)

d <- read.csv("content/tessera/data/csv/iris.csv")
d$Species <- factor(
  d$Species,
  levels = c("setosa", "versicolor", "virginica")
)

walter_white2 <- get_palette("walter_white2", type = "qualitative")
species_colors <- setNames(
  walter_white2[c(1, 2, 4)],
  levels(d$Species)
)
species_shapes <- setNames(c(21, 22, 24), levels(d$Species))

scatter_scales <- list(
  scale_x_continuous(
    breaks = 1:7,
    limits = c(0.9, 7.1),
    expand = expansion(mult = c(0, 0))
  ),
  scale_y_continuous(
    breaks = seq(0, 2.5, 0.5),
    limits = c(-0.05, 2.55),
    expand = expansion(mult = c(0, 0))
  )
)

scatter_theme <- theme_pubr(base_size = 13, legend = "right") +
  theme(
    panel.grid.major = element_line(color = "#E5E3DC", linewidth = 0.35),
    panel.grid.minor = element_blank(),
    axis.line = element_line(color = "#333330", linewidth = 0.45),
    axis.ticks = element_line(color = "#333330", linewidth = 0.4),
    plot.title = element_text(face = "bold", size = 15, hjust = 0),
    legend.title = element_text(face = "bold"),
    legend.key.height = grid::unit(5, "mm"),
    strip.background = element_blank(),
    strip.text = element_text(face = "bold"),
    plot.margin = margin(14, 16, 12, 12)
  )

A · 普通散点

没有分组要表达时，只画点。不要为了“丰富”画面主动加入颜色或形状。

#| fig: basic
#| fig-width: 7.4
#| fig-height: 5.1
ggplot(d, aes(Petal.Length, Petal.Width)) +
  geom_point(size = 2.7, alpha = 0.72, color = "#333330") +
  scatter_scales +
  labs(
    title = "Iris petal measurements",
    x = "Petal length (cm)",
    y = "Petal width (cm)"
  ) +
  scatter_theme

B · 用形状分组

形状适合黑白输出，也适合颜色不能承担信息的场景。三个物种的点云聚集位置不同，因此即使没有颜色也能读出分组结构。

#| fig: shape
#| fig-width: 7.4
#| fig-height: 5.1
ggplot(d, aes(Petal.Length, Petal.Width, shape = Species)) +
  geom_point(size = 3.1, alpha = 0.72, color = "#333330", stroke = 0.7) +
  scale_shape_manual(values = species_shapes) +
  scatter_scales +
  labs(
    title = "Grouped by shape",
    x = "Petal length (cm)",
    y = "Petal width (cm)",
    shape = "Species"
  ) +
  scatter_theme

C · 用颜色分组

这是彩色输出下最直接的分组方案。所有点保持同一种圆形，读者只需要解一套颜色图例；同色点在空间上形成点群，分组关系会比 Orange 示例清楚得多。

#| fig: color
#| fig-width: 7.4
#| fig-height: 5.1
ggplot(d, aes(Petal.Length, Petal.Width, fill = Species)) +
  geom_point(
    shape = 21,
    size = 3.2,
    alpha = 0.82,
    color = "#333330",
    stroke = 0.55
  ) +
  scale_fill_manual(values = species_colors) +
  scatter_scales +
  labs(
    title = "Grouped by color",
    x = "Petal length (cm)",
    y = "Petal width (cm)",
    fill = "Species"
  ) +
  scatter_theme

D · 形状与颜色共同分组

两种通道必须映射同一个变量。这样没有增加新的信息，只是让分组在色觉差异、灰度打印或投影偏色时仍然可辨。普通屏幕展示优先使用 C；确实需要冗余编码时再用这一版。

#| fig: color-shape
#| fig-width: 7.4
#| fig-height: 5.1
ggplot(d, aes(Petal.Length, Petal.Width, fill = Species, shape = Species)) +
  geom_point(
    size = 3.2,
    alpha = 0.82,
    color = "#333330",
    stroke = 0.6
  ) +
  scale_fill_manual(values = species_colors) +
  scale_shape_manual(values = species_shapes) +
  scatter_scales +
  labs(
    title = "Color + shape",
    x = "Petal length (cm)",
    y = "Petal width (cm)",
    fill = "Species",
    shape = "Species"
  ) +
  scatter_theme

E · 调整点大小

size 控制的是点的直径感，而不是简单的装饰。点太小，缩放后消失；点太大，组内密集区域会糊在一起。下面复制同一份数据做三个分面，只改变 geom_point() 的大小，颜色、透明度和坐标完全相同。

#| fig: point-size
#| fig-width: 11
#| fig-height: 4.2
size_settings <- data.frame(
  size_label = factor(
    c("size = 1.5", "size = 3", "size = 5"),
    levels = c("size = 1.5", "size = 3", "size = 5")
  ),
  point_size = c(1.5, 3, 5)
)

size_demo <- merge(d, size_settings)

ggplot(
  size_demo,
  aes(Petal.Length, Petal.Width, fill = Species, size = point_size)
) +
  geom_point(shape = 21, alpha = 0.78, color = "#333330", stroke = 0.5) +
  facet_wrap(~size_label, nrow = 1) +
  scale_fill_manual(values = species_colors) +
  scale_size_identity() +
  scatter_scales +
  guides(fill = "none") +
  labs(
    title = "Point size changes overlap and visual weight",
    x = "Petal length (cm)",
    y = "Petal width (cm)"
  ) +
  scatter_theme +
  theme(plot.title = element_text(size = 14))

这里 size = 3 是这份数据和当前画布的平衡点，不是通用默认值。样本量、重叠程度、输出尺寸和导出后的缩放比例变了，都应重新判断。

五种方法怎么选

1. 没有分组，选 A。
2. 黑白输出或不能使用颜色，选 B。
3. 常规彩色图，选 C。
4. 需要无障碍或灰度兼容，才选 D；不要把颜色和形状分别映射到两个不同变量。
5. 最后根据点数和最终版面调整 size；不要把预览窗口里的默认值当成成品参数。