如何用ggplot2绘制大量折线并有效凸显重叠区域？

我正在开展一项实验，需对两个模型进行数百次拟合，希望绘制所有运行结果的折线图——不同颜色代表不同模型，单条折线代表一次运行。我使用ggplot2的geom_line()函数，尝试调整alpha参数让单条折线变淡、重叠区域加深以凸显“主线”，但并未奏效，推测是多数折线实际并未重叠。请问该如何改进？

当前代码如下：

modelB_color <- rgb(0, 0, 0, alpha = alpha_value)
modelA_color <- rgb(1, 0, 0, alpha = alpha_value)
for (data in simulation_data) {
        modelA <- data$modelA
        modelB <- data$modelB

        x <- seq(0, 1, length = 100)

        g <- g + geom_line(data = modelB, aes(x = x, y = y_probs, color = "modelA"), linewidth = 0.5)
        g <- g + geom_line(data = modelA, aes(x = x, y = y, color = "modelA"), linewidth = 0.5)
}
g <- g + scale_color_manual(
        name = "Model",
        values = c("modelB" = modelB_color, "modelA" = modelA_color),
        labels = c("modelB", "modelA")
    )

当前效果如图：

1. 重构数据格式（ggplot2最优实践） #

ggplot2原生适配长格式数据，循环叠加图层不仅效率低，还容易出错。先把所有模拟数据合并为统一数据框，包含模型类型、x、y、运行编号字段：

# 初始化空数据框
all_data <- data.frame(
  model = character(),
  x = numeric(),
  y = numeric(),
  run_id = integer()
)

# 遍历模拟结果并合并数据
for (i in seq_along(simulation_data)) {
  data <- simulation_data[[i]]
  # 处理Model A数据
  modelA_df <- data.frame(
    model = "modelA",
    x = seq(0, 1, length = 100),
    y = data$modelA$y,
    run_id = i
  )
  # 处理Model B数据
  modelB_df <- data.frame(
    model = "modelB",
    x = seq(0, 1, length = 100),
    y = data$modelB$y_probs,
    run_id = i
  )
  # 合并到总数据框
  all_data <- rbind(all_data, modelA_df, modelB_df)
}

2. 用分位数/密度凸显核心趋势 #

如果折线重叠度低，alpha叠加效果不明显，可通过以下方式强化趋势：

• 分位数线+区间填充：直接展示每条x对应的中位数、四分位区间，清晰凸显核心趋势：

library(dplyr)

# 计算分位数统计量
quantile_data <- all_data %>%
  group_by(model, x) %>%
  summarize(
    median_y = median(y),
    q25_y = quantile(y, 0.25),
    q75_y = quantile(y, 0.75)
  )

# 绘图：叠加所有折线+分位数线+区间填充
ggplot(all_data, aes(x = x, y = y, color = model)) +
  geom_line(aes(group = run_id), alpha = 0.1, linewidth = 0.3) + # 单条折线调淡
  geom_line(data = quantile_data, aes(y = median_y), linewidth = 1.2) + # 中位数线加粗
  geom_ribbon(data = quantile_data, aes(ymin = q25_y, ymax = q75_y, fill = model), alpha = 0.2) + # 四分位区间填充
  scale_color_manual(values = c("modelA" = "red", "modelB" = "black")) +
  scale_fill_manual(values = c("modelA" = "red", "modelB" = "black")) +
  labs(title = "模型模拟结果趋势", x = "X轴", y = "Y值", color = "模型", fill = "模型")

• 2D密度热力图：用密度色块展示数据集中区域，替代单纯的折线叠加：

ggplot(all_data, aes(x = x, y = y, color = model)) +
  stat_density_2d(aes(fill = model), geom = "polygon", alpha = 0.3, bins = 15) +
  geom_line(aes(group = run_id), alpha = 0.05, linewidth = 0.2) +
  scale_color_manual(values = c("modelA" = "red", "modelB" = "black")) +
  scale_fill_manual(values = c("modelA" = "red", "modelB" = "black"))

3. 修复原代码的核心错误 #

原代码存在映射错误：Model B的折线被错误标记为modelA，导致颜色区分失效。修正后的代码（仍不推荐循环图层，仅作错误修复参考）：

alpha_value <- 0.1
modelB_color <- rgb(0, 0, 0, alpha = alpha_value)
modelA_color <- rgb(1, 0, 0, alpha = alpha_value)

g <- ggplot() # 初始化ggplot对象
for (data in simulation_data) {
        modelA <- data$modelA
        modelB <- data$modelB

        x <- seq(0, 1, length = 100)

        # 修正color映射，Model B对应"modelB"
        g <- g + geom_line(data = modelB, aes(x = x, y = y_probs, color = "modelB"), linewidth = 0.5)
        g <- g + geom_line(data = modelA, aes(x = x, y = y, color = "modelA"), linewidth = 0.5)
}
g <- g + scale_color_manual(
        name = "Model",
        values = c("modelB" = modelB_color, "modelA" = modelA_color),
        labels = c("modelB", "modelA")
    )

内容的提问来源于stack exchange，提问作者Ozay0900