如何在ggsankey桑基图中兼顾正确标签与流大小？

问题根源在于原数据中同一节点对应多条流向记录：当映射value时，geom_sankey能正确计算流量，但geom_sankey_label会为每条记录生成标签，导致重复；去掉value映射后标签正常，但流量计算失效。

解决步骤： #

1. 提取唯一节点信息：从原数据中筛选出每个阶段（x）与节点（node）的唯一组合，确保每个节点仅对应一条记录。
2. 绘图时，主数据流保留value映射保证流量正确，给geom_sankey_label单独指定唯一节点数据集，避免标签重复。

完整代码： #

# Import packages
library(tidyverse)
library(ggsankey)

# create data
test <- tibble(
  x = factor(c("Production", "Production", "Production", "Production", "Production", "Production", "Transport", "Transport", "Consumption", "Consumption"),
             levels = c("Production", "Transport", "Consumption")),
  node = c("Product A", "Product B", "Product A", "Product B", "Product A", "Product B", "Transport", "Transport", "Sector I", "Sector II"),
  next_x = factor(c("Transport", "Transport", "Consumption", "Consumption", "Consumption", "Consumption", "Consumption", "Consumption", NA_character_, NA_character_),
                  levels = c("Production", "Transport", "Consumption", NA_character_)),
  next_node = c("Transport", "Transport", "Sector I", "Sector I", "Sector II", "Sector II", "Sector I", "Sector II", NA_character_, NA_character_),
  value = c(1323, 1159, 5070, 2554, 461, 2758, 1768, 713, 9392, 3932)
)

# 提取唯一节点数据集（每个x+node组合仅保留一条）
node_labels <- test %>%
  distinct(x, node) %>%
  drop_na()

# 绘制桑基图
ggplot(test, aes(x = x,
                 next_x = next_x,
                 node = node,
                 next_node = next_node,
                 value = value,
                 fill = factor(node))) +
  geom_sankey(flow.alpha = .6, color = "gray30") +
  # 单独指定标签用的数据集，确保每个节点只显示一次
  geom_sankey_label(data = node_labels, 
                    aes(label = node),
                    size = 3, color = "white", fill = "gray40") +
  scale_fill_viridis_d(drop = FALSE) +
  theme_sankey(base_size = 18) +
  labs(x = NULL) +
  theme(legend.position = "none",
        axis.text.x = element_blank(),
        plot.title = element_text(hjust = .5)) +
  ggtitle("Massflow")

效果说明： #

• 主数据集test保留value映射，geom_sankey会根据每条记录的value计算流的宽度，流量大小完全匹配数据。
• geom_sankey_label使用node_labels数据集，每个节点在对应阶段仅绘制一次标签，解决重复问题。

内容的提问来源于stack exchange，提问作者Hugo