如何用dplyr创建多变量占比表？含R基础实现对比与占比需求

原始数据 #

A <- rep(c("A1", "A2", "A3"), 10)
B <- sample(1:5, size=30, replace=TRUE)
C <- sample(c("yes", "no"), size=30, replace=TRUE)

问题1：用tidyverse的dplyr结合group_by、summarize实现目标表格 #

要复刻基础Rdata.frame(rbind(table(B, A), table(C, A)))的输出效果，可按长格式转换→分组统计→宽格式重塑的流程实现：

library(tidyverse)

# 整合数据并转成长格式，统一处理B、C变量
df <- tibble(A, B, C) %>%
  pivot_longer(cols = c(B, C), names_to = "variable", values_to = "value")

# 分组计数后转宽格式，得到目标表格
result_table <- df %>%
  group_by(variable, value, A) %>%
  summarize(count = n(), .groups = "drop") %>%
  pivot_wider(names_from = A, values_from = count, values_fill = 0)

print(result_table)

核心逻辑：

• pivot_longer将B、C两个独立变量合并为variable（标记变量类型）和value（变量取值）两列，便于统一分组统计。
• group_by(variable, value, A)按变量类型、变量取值、A的类别分组，summarize(count = n())统计每组样本量。
• pivot_wider将A的类别（A1/A2/A3）转为列，values_fill = 0确保无数据的单元格填充0，和基础R的table输出逻辑一致。

问题2：为单元格添加占比 #

以下提供两种常见占比的实现方式，以行占比（对应variable-value行内的占比）为例：

result_table_with_row_pct <- df %>%
  group_by(variable, value, A) %>%
  summarize(count = n(), .groups = "drop") %>%
  group_by(variable, value) %>%
  mutate(pct = scales::percent(count / sum(count))) %>% # 格式化百分比，依赖scales包
  mutate(cell = paste0(count, " (", pct, ")")) %>% # 合并计数与占比
  select(-count, -pct) %>%
  pivot_wider(names_from = A, values_from = cell, values_fill = "0 (0%)")

print(result_table_with_row_pct)

若需要列占比（对应A列内的占比），仅需调整分组逻辑：

result_table_with_col_pct <- df %>%
  group_by(variable, value, A) %>%
  summarize(count = n(), .groups = "drop") %>%
  group_by(variable, A) %>%
  mutate(pct = scales::percent(count / sum(count))) %>%
  mutate(cell = paste0(count, " (", pct, ")")) %>%
  select(-count, -pct) %>%
  pivot_wider(names_from = A, values_from = cell, values_fill = "0 (0%)")

print(result_table_with_col_pct)

注：若不想依赖scales包，可手动格式化百分比：paste0(round(count / sum(count)*100, 1), "%")。

内容的提问来源于stack exchange，提问作者Produnis