如何使用R语言ggplot包可视化目标向量与数据框各列的相关系数

首先需要把计算得到的宽格式相关系数结果转换为ggplot支持的长数据格式，以下代码可直接复现全流程：

library(tidyverse)

# --------------- 数据准备（你已完成的步骤，可直接替换为真实数据）---------------
set.seed(123) # 仅用于结果复现，跑真实数据时可删除
y1 = rpois(10,3) 
y2 = rpois(10,5)
y3 = rpois(10,10)
y4 = rpois(10,20)
data = tibble(y1,y2,y3,y4)
xx  = rpois(10,4)

# 计算相关系数并转长格式，是绘图的必要前置步骤
cor_df <- data %>%
  dplyr::summarize(across(.cols = everything(), ~ cor(.x,xx, use = "pairwise.complete.obs"))) %>%
  pivot_longer(cols = everything(), names_to = "var_name", values_to = "cor_coef")

以下提供3种常用的可视化方案，所有方案都固定相关系数色阶范围为[-1,1]，避免色阶自适应导致的视觉误判。

方案1：水平条形图（可读性最高，首推） #

正负相关方向、数值大小辨识度最高，适合正式分析报告使用：

ggplot(cor_df, aes(x = reorder(var_name, cor_coef), y = cor_coef, fill = cor_coef)) +
  geom_col(width = 0.65) +
  # 直接在条形上标注相关系数数值，自动适配正负方向的标注位置
  geom_text(aes(label = round(cor_coef, 3),
                hjust = ifelse(cor_coef >= 0, -0.2, 1.2)),
            size = 4) +
  # 用红-白-蓝渐变区分负相关、零相关、正相关，符合统计可视化惯例
  scale_fill_gradient2(low = "#d73027", mid = "white", high = "#4575b4",
                       midpoint = 0, limits = c(-1, 1), name = "相关系数") +
  geom_hline(yintercept = 0, color = "grey30", linewidth = 0.8) +
  coord_flip() +
  labs(x = "数据框变量", y = "与目标向量xx的相关系数", title = "各变量与xx的相关系数") +
  theme_minimal() +
  theme(plot.title = element_text(hjust = 0.5))

方案2：单行热图 #

适合和其他多变量热图拼接、统一排版使用：

ggplot(cor_df, aes(x = var_name, y = "相关系数", fill = cor_coef)) +
  geom_tile(color = "white", linewidth = 1) +
  geom_text(aes(label = round(cor_coef, 3)), size = 4) +
  scale_fill_gradient2(low = "#d73027", mid = "white", high = "#4575b4",
                       midpoint = 0, limits = c(-1, 1), name = "相关系数") +
  labs(x = "变量", y = NULL, title = "相关系数热图") +
  theme_minimal() +
  theme(plot.title = element_text(hjust = 0.5),
        panel.grid = element_blank())

方案3：棒棒糖散点图 #

样式更简洁，点的大小和颜色同时映射相关系数大小：

ggplot(cor_df, aes(x = var_name, y = cor_coef, color = cor_coef, size = abs(cor_coef))) +
  geom_segment(aes(x = var_name, xend = var_name, y = 0, yend = cor_coef), color = "grey60") +
  geom_point() +
  geom_text(aes(label = round(cor_coef, 3), 
                vjust = ifelse(cor_coef >= 0, -1, 1.5)),
            size = 3.5, color = "black") +
  scale_color_gradient2(low = "#d73027", mid = "white", high = "#4575b4",
                       midpoint = 0, limits = c(-1, 1)) +
  geom_hline(yintercept = 0, color = "grey30", linewidth = 0.8) +
  scale_size_continuous(range = c(2,6)) +
  labs(x = "变量", y = "与xx的相关系数", title = "相关系数散点图") +
  theme_minimal() +
  theme(plot.title = element_text(hjust = 0.5),
        legend.position = "none")

小提示：如果需要调整配色、字体、标注样式，直接修改对应geom的参数即可。

内容的提问来源于stack exchange，提问作者Homer Jay Simpson