如何按通过结果、关联关系同时统计事件数与人数生成汇总表？

核心思路是无需拆分多次计算，通过一次分组聚合生成多维度统计结果，再通过长宽转换直接得到目标矩阵，省略手动填值的冗余步骤：

# 加载依赖包
library(tidyverse)

# 统一设置Pass因子水平，避免多次重复定义，同时保留NA分组
df <- df %>% 
  mutate(Pass = factor(Pass, levels = c("N", "Y", NA), exclude = NULL))

# 1. 按Relation+Pass分组统计事件数
event_stats <- df %>%
  group_by(Relation, Pass, .drop = FALSE) %>%
  summarise(event_count = n(), .groups = "drop")

# 2. 统计分Relation+Pass的去重人数，同时追加全量不分Relation的人数统计
people_stats <- df %>%
  group_by(Relation, Pass, .drop = FALSE) %>%
  summarise(people_count = n_distinct(ID), .groups = "drop") %>%
  bind_rows(
    df %>%
      group_by(Pass, .drop = FALSE) %>%
      summarise(people_count = n_distinct(ID), .groups = "drop") %>%
      mutate(Relation = "ALL")
  )

# 3. 合并结果转为宽表，直接输出整理好的汇总矩阵
result <- event_stats %>%
  left_join(people_stats, by = c("Relation", "Pass")) %>%
  pivot_wider(
    id_cols = Pass,
    names_from = Relation,
    values_from = c(event_count, people_count)
  )

代码优化了你原实现中的冗余操作，也修正了events_rl部分filter语句缺少闭合括号的语法问题。运行后得到的result对象已经包含所有你需要的统计维度数据，无需手动拼接填值。

内容的提问来源于stack exchange，提问作者Stataq