R语言冲积图实现所有Stratum尺寸一致的技术问询

我用R语言绘制包含两个stratum的冲积图，两个stratum拥有相同类别，但因观测数量不同，绘图后各类别的尺寸存在差异，希望实现所有stratum尺寸一致的效果。

原绘图代码：

g_36a_flow <- ggplot(t_14_long, 
                     aes(alluvium = oper_num, x = class, 
                         stratum = factor(rating_qal), 
                         fill = factor(rating_qal))) + 
  geom_flow(position = position_dodge(width = 0), 
            aes(fill = factor(rating_qal)), width = 0.2) +
  geom_stratum(width = 0.1, color = "white") +
  geom_text(data = t_14_long, 
            aes(x = as.numeric(factor(class)) + .075 * ifelse(class == "MGMT", 1.5, -1.5),
                y = after_stat(y),    
                label = paste0(label_num, "%"),
                hjust = ifelse(class == "MGMT", 1, 0)),
            stat = "stratum", size = 10, color = "black", family = "montserrat") +
  geom_text(aes(label = after_stat(stratum),
                hjust = ifelse(class == "MGMT", 1, 0),
                x = as.numeric(factor(class)) + .075 * ifelse(class == "MGMT", -1, 1),),
            stat = "stratum", size = 10, color = "black", family = "montserrat"  ) +
  ggtitle("g_36a_flow: Management vs OVE: % of projects by rating
validation cycle 2018-2023 ") +
  scale_fill_manual(values = c("#78B743", "#A9CD74", "#DBE9AA", "#F9BEC4", "#F4817D", "#EF3648"),
                    breaks = levels(t_14_long$rating_qal_aux),
                    labels = levels(t_14_long$rating_qal_aux)) +
  labs(x = "", y = "") +
  theme_void() + 
  theme(text       = element_text(family = "montserrat", 
                                  size = 30, color = "black"),
        axis.text  = element_text(size = 30, color = "black"),
        axis.title = element_text(size = 30, color = "black"),
        plot.title = element_text(size = 30, color = "black"),
        legend.position = "none",
        axis.title.y = element_text(angle = 90, vjust = 0.5),
        plot.background = element_rect(fill = "white"),
        panel.border = element_blank()) +
  scale_x_discrete(expand = c(0.1, 0.1)) +
  scale_y_discrete(expand = c(0.1, 0.1)) 

g_36a_flow

解决方案 #

要让两个stratum的每个类别尺寸一致，核心是让每个类别的权重相同，忽略实际观测数量的影响：

1. 先给数据添加权重列，确保每个class和rating_qal的组合权重一致：

library(dplyr)

t_14_long <- t_14_long %>%
  group_by(class, rating_qal) %>%
  mutate(stratum_weight = 1) %>% # 每个组合权重设为1
  ungroup()

2. 修改ggplot的映射，加入weight = stratum_weight参数，强制每个stratum按相同权重计算高度：

g_36a_flow <- ggplot(t_14_long, 
                     aes(alluvium = oper_num, x = class, 
                         stratum = factor(rating_qal), 
                         fill = factor(rating_qal),
                         weight = stratum_weight)) + # 新增权重映射
  geom_flow(position = position_dodge(width = 0), 
            aes(fill = factor(rating_qal)), width = 0.2) +
  geom_stratum(width = 0.1, color = "white") +
  # 以下文本标注代码保持不变
  geom_text(data = t_14_long, 
            aes(x = as.numeric(factor(class)) + .075 * ifelse(class == "MGMT", 1.5, -1.5),
                y = after_stat(y),    
                label = paste0(label_num, "%"),
                hjust = ifelse(class == "MGMT", 1, 0)),
            stat = "stratum", size = 10, color = "black", family = "montserrat") +
  geom_text(aes(label = after_stat(stratum),
                hjust = ifelse(class == "MGMT", 1, 0),
                x = as.numeric(factor(class)) + .075 * ifelse(class == "MGMT", -1, 1),),
            stat = "stratum", size = 10, color = "black", family = "montserrat"  ) +
  ggtitle("g_36a_flow: Management vs OVE: % of projects by rating
validation cycle 2018-2023 ") +
  scale_fill_manual(values = c("#78B743", "#A9CD74", "#DBE9AA", "#F9BEC4", "#F4817D", "#EF3648"),
                    breaks = levels(t_14_long$rating_qal_aux),
                    labels = levels(t_14_long$rating_qal_aux)) +
  labs(x = "", y = "") +
  theme_void() + 
  theme(text       = element_text(family = "montserrat", 
                                  size = 30, color = "black"),
        axis.text  = element_text(size = 30, color = "black"),
        axis.title = element_text(size = 30, color = "black"),
        plot.title = element_text(size = 30, color = "black"),
        legend.position = "none",
        axis.title.y = element_text(angle = 90, vjust = 0.5),
        plot.background = element_rect(fill = "white"),
        panel.border = element_blank()) +
  scale_x_discrete(expand = c(0.1, 0.1)) +
  scale_y_discrete(expand = c(0.1, 0.1)) 

g_36a_flow

这样修改后，两个stratum的每个类别高度会保持一致，达到预期效果。

内容的提问来源于stack exchange，提问作者Luis Fernando Corrales