如何修改rms包绘图中的线型？scale_linetype_manual失效问题

核心遗漏点 #

你遇到的问题本质是生存分析绘图时未将分组变量映射到linetype美学属性，导致scale_linetype_manual没有生效的目标对象。以下结合模拟代码拆解差异：

生效的lm模型代码逻辑 #

lm绘图时，你已经在aes()里明确将分组变量绑定到linetype，所以后续的线型设置能精准作用：

library(ggplot2)
set.seed(123)
df <- data.frame(
  x = rnorm(100),
  y = rnorm(100),
  group = factor(sample(c("A", "B", "C"), 100, replace = TRUE))
)
model_lm <- lm(y ~ x * group, data = df)
pred_df_lm <- predict(model_lm, newdata = expand.grid(x = seq(-2,2,0.1), group = c("A", "B", "C")), se.fit = TRUE) |>
  (\(x) data.frame(x = seq(-2,2,0.1), group = rep(c("A", "B", "C"), each = 41), fit = x$fit))()

# 关键：aes中声明linetype = group
ggplot(pred_df_lm, aes(x = x, y = fit, linetype = group)) +
  geom_line() +
  scale_linetype_manual(values = c("solid", "dashed", "dashed"))

生存分析绘图的问题所在 #

rms的Predict返回生存数据时，你大概率只映射了color到分组，没给linetype绑定分组变量，导致线型设置无的放矢：

library(rms)
library(survival)
set.seed(123)
surv_df <- data.frame(
  time = rexp(200),
  status = sample(c(0,1), 200, replace = TRUE),
  group = factor(sample(c("A", "B", "C"), 200, replace = TRUE))
)
model_cph <- cph(Surv(time, status) ~ group, data = surv_df, surv = TRUE)
pred_cph <- Predict(model_cph, group = c("A", "B", "C"), type = "survival")

# 错误：仅映射color，未映射linetype
ggplot(pred_cph, aes(x = time, y = survival)) +
  geom_step(aes(color = group)) +
  scale_linetype_manual(values = c("solid", "dashed", "dashed"))

修正后的代码 #

只需在aes()中添加linetype = group的映射即可，同时可保留颜色映射增强区分度：

ggplot(pred_cph, aes(x = time, y = survival, linetype = group, color = group)) +
  geom_step() +
  scale_linetype_manual(values = c("solid", "dashed", "dashed")) +
  # 可选：自定义分组颜色
  scale_color_manual(values = c("#000000", "#E69F00", "#56B4E9"))

额外注意事项 #

1. 确保分组变量是因子类型：如果pred_cph$group是字符型，可提前转换为因子避免排序混乱：

   pred_cph$group <- factor(pred_cph$group, levels = c("A", "B", "C"))
   

2. 线型值数量需与分组数匹配：如果有3个分组，values参数必须提供3个线型字符串。

内容的提问来源于stack exchange，提问作者li jiaqi