如何在ggplot的geom_smooth中按回归线斜率设置颜色或线条粗细

实现方法 #

你需要先预先计算每个受试者对应的回归斜率，再将斜率映射为ggplot的颜色或线条大小美学即可，具体操作如下：

步骤1：计算每位受试者的回归斜率 #

先对数据集按ID分组，拟合每个ID的线性模型并提取斜率，再将斜率匹配回原始数据集：

library(dplyr)

# 按ID分组计算回归斜率
slope_data <- mydata %>%
  group_by(ID) %>%
  summarise(lm_slope = coef(lm(continuous_outcome ~ age, data = cur_data()))[[2]])

# 合并斜率到原数据集
mydata_new <- left_join(mydata, slope_data, by = "ID")

步骤2：按斜率调整可视化效果 #

你可以按需选择以下任意一种或组合实现效果：

• 效果1：斜率越高颜色越偏红

ggplot(mydata_new, aes(x = age, y = continuous_outcome, group = ID)) +
  geom_smooth(aes(color = lm_slope), method = "lm", formula = y~x, se = FALSE, size = 0.5) +
  # 颜色渐变：低斜率为浅蓝色，高斜率为深红色
  scale_color_gradient(low = "lightblue", high = "darkred") +
  theme(legend.position = "none") # 不需要图例就保留这句，需要看斜率对应值就删掉

• 效果2：斜率越高线条越粗

ggplot(mydata_new, aes(x = age, y = continuous_outcome, group = ID)) +
  geom_smooth(aes(size = lm_slope), method = "lm", formula = y~x, se = FALSE, color = "gray30") +
  # 调整线条粗细范围，可按需修改
  scale_size_continuous(range = c(0.2, 2)) +
  theme(legend.position = "none")

• 效果3：同时调整颜色和粗细，识别度更高

ggplot(mydata_new, aes(x = age, y = continuous_outcome, group = ID)) +
  geom_smooth(aes(color = lm_slope, size = lm_slope), method = "lm", formula = y~x, se = FALSE) +
  scale_color_gradient(low = "lightblue", high = "darkred") +
  scale_size_continuous(range = c(0.2, 2)) +
  theme(legend.position = "none")

可选优化：区分正负斜率 #

如果你的数据中存在斜率为负（随年龄增长结局指标下降）的情况，可以使用发散色盘更清晰区分升降趋势：

ggplot(mydata_new, aes(x = age, y = continuous_outcome, group = ID)) +
  geom_smooth(aes(color = lm_slope, size = abs(lm_slope)), method = "lm", formula = y~x, se = FALSE) +
  # 负斜率为蓝色，0附近为白色，正斜率为红色
  scale_color_gradient2(low = "darkblue", mid = "white", high = "darkred", midpoint = 0) +
  scale_size_continuous(range = c(0.2, 2)) +
  theme(legend.position = "none")

内容的提问来源于stack exchange，提问作者tcvdb1992