ggplot按x轴total_meeting值重排y轴bizdev失效问题求助

背景数据与处理流程 #

原始数据集结构：

df <- tibble(
  bizdev = c("Jerry Harding", "Steve Vargas", "Steve Vargas", "Jerry Harding", "Jerry Harding", "David Burton"),
  company = c("Jones, Figueroa and Moon", "Stanley, Porter and Mccoy", "Gardner and Sons", "Parks-Chen", "Bond Group", "Lopez-Johnson"),
  date = as.Date(c("2020-04-06", "2020-04-24", "2020-04-19", "2020-04-11", "2020-04-06", "2020-05-07")),
  category = c("Mid-tier", "Indie", "Indie", "Indie", "Indie", "Enterprise")
)

数据转换代码：

consol <- df %>% 
  mutate(meeting_month = as_factor(month(date, abbr = TRUE, label = TRUE)), .keep = 'unused') %>%
  group_by(bizdev, category, meeting_month) %>% 
  summarise(total_meeting = n(), .groups = "drop_last")

转换后数据集（部分示例）：

# Groups:   bizdev, category [22]
   bizdev       category   meeting_month total_meeting
   <chr>        <chr>      <ord>                 <int>
 1 Bobby Adams  Enterprise Apr                       2
 2 Bobby Adams  Indie      Mar                      12
 3 Bobby Adams  Indie      Apr                      13
 4 Bobby Adams  Indie      May                      12
 5 Bobby Adams  Mid-tier   Mar                       6
 6 Bobby Adams  Mid-tier   Apr                       6
 7 Bobby Adams  Mid-tier   May                       3
 8 David Burton Enterprise May                       2
 9 David Burton Indie      Mar                      30
10 David Burton Indie      Apr                      31

问题现象 #

使用以下代码绘图时，y轴的bizdev未按total_meeting的预期值排序（比如Steve Vargas的total_meeting高于Joel English，但位置错误）：

consol %>% 
  ggplot(aes(x = total_meeting, y = reorder(bizdev, -total_meeting), fill = category)) +
  geom_bar(stat = 'identity') +
  theme_classic()

原因分析 #

reorder(bizdev, -total_meeting)默认会对每个bizdev对应的所有total_meeting值取算术平均值作为排序依据，而非你需要的总会议数、最大单月会议数等汇总值。当一个bizdev同时存在高值和低值记录时，均值会拉平排序权重，导致显示顺序不符合预期。

解决方案 #

方案1：按每个bizdev的总会议数排序

先计算每个业务开发人员的累计会议总数，再基于该值排序y轴：

# 计算每个bizdev的总会议数
bizdev_total <- consol %>% 
  group_by(bizdev) %>% 
  summarise(total_all = sum(total_meeting)) %>% 
  ungroup()

# 合并数据并绘图
consol %>% 
  left_join(bizdev_total, by = "bizdev") %>% 
  ggplot(aes(x = total_meeting, y = reorder(bizdev, -total_all), fill = category)) +
  geom_col() + # geom_col是geom_bar(stat='identity')的简写，更直观
  theme_classic()

方案2：按每个bizdev的最大单月会议数排序

如果需要按单月最高会议数排序：

# 计算每个bizdev的最大单月会议数
bizdev_max <- consol %>% 
  group_by(bizdev) %>% 
  summarise(max_meeting = max(total_meeting)) %>% 
  ungroup()

# 合并数据并绘图
consol %>% 
  left_join(bizdev_max, by = "bizdev") %>% 
  ggplot(aes(x = total_meeting, y = reorder(bizdev, -max_meeting), fill = category)) +
  geom_col() +
  theme_classic()

方案3：用forcats包的fct_reorder2直接排序

借助forcats包的fct_reorder2函数，可直接在ggplot中基于指定规则排序，无需提前合并数据：

library(forcats)

consol %>% 
  ggplot(aes(x = total_meeting, 
             y = fct_reorder2(bizdev, total_meeting, category, .fun = max), # 按每个bizdev的最大total_meeting排序
             fill = category)) +
  geom_col() +
  theme_classic()

内容的提问来源于stack exchange，提问作者Ong K.S