如何为同一数据生成的叠加散点图与线图手动指定不同颜色？

要让散点图和线图使用不同逻辑的颜色区分，同时保留分组信息，有两种常用实现方式：

方法1：手动指定颜色并自定义图例 #

直接在geom_point和geom_line中分别设置颜色映射，再通过scale_color_manual定义对应颜色，让图例清晰区分实际值（散点）与预测值（线）：

set.seed(123)
df2 <- data.frame(
  time = rep(1:10, 2),
  value = c(runif(10), runif(10) + 1),
  predicted_value = c(runif(10), runif(10) + 1.5),
  youngOld = rep(c("young", "old"), each = 10)
)

library(ggplot2)

sample <- ggplot(df2, aes(x = time)) + 
  # 散点映射为「分组_实际值」的标识
  geom_point(aes(y = value, color = paste(youngOld, "actual")), size = 2) +
  # 线映射为「分组_预测值」的标识
  geom_line(aes(y = predicted_value, color = paste(youngOld, "predicted")), linewidth = 1.2) +
  # 手动指定每组颜色，同分组的散点和线颜色一致
  scale_color_manual(
    values = c(
      "young actual" = "#1f77b4", 
      "old actual" = "#ff7f0e", 
      "young predicted" = "#1f77b4", 
      "old predicted" = "#ff7f0e"
    ),
    # 自定义图例符号：实际值显示点，预测值显示线
    guide = guide_legend(override.aes = list(
      shape = c(16, 16, NA, NA),
      linetype = c(0, 0, 1, 1)
    ))
  ) +
  labs(title = "实际值与预测值对比",
       x = "时间",
       y = "数值",
       color = "分组与数据类型")
sample

如果只是想解决遮挡问题，也可以直接把geom_line代码放在geom_point之后，让线显示在散点上方，无需修改颜色逻辑。

方法2：整理数据为长格式（更推荐） #

用tidyr包将实际值、预测值合并到同一列，新增标记列区分数据类型，更贴合ggplot的语法逻辑：

set.seed(123)
df2 <- data.frame(
  time = rep(1:10, 2),
  value = c(runif(10), runif(10) + 1),
  predicted_value = c(runif(10), runif(10) + 1.5),
  youngOld = rep(c("young", "old"), each = 10)
)

library(ggplot2)
library(tidyr)

# 转换为长格式数据
df_long <- df2 %>%
  pivot_longer(cols = c(value, predicted_value),
               names_to = "data_type",
               values_to = "num_value") %>%
  mutate(data_type = ifelse(data_type == "value", "实际值", "预测值"))

# 绘图：用颜色映射分组，线型映射数据类型
sample <- ggplot(df_long, aes(x = time, y = num_value)) +
  geom_point(aes(color = youngOld), data = subset(df_long, data_type == "实际值"), size = 2) +
  geom_line(aes(color = youngOld, linetype = data_type), data = subset(df_long, data_type == "预测值"), linewidth = 1.2) +
  scale_linetype_manual(values = c("预测值" = 1)) +
  labs(title = "实际值与预测值对比",
       x = "时间",
       y = "数值",
       color = "分组",
       linetype = "数据类型")
sample

这种方式的数据结构更清晰，后续调整样式、扩展分析更灵活。

内容的提问来源于stack exchange，提问作者Nereus