使用dplyr为数据框补全cluster 1-10的缺失treatment行

原始数据 #

df <- structure(list(cluster = c(2L, 3L, 5L, 5L, 6L, 6L, 7L, 9L, 9L, 
10L, 10L), treatment = c("TreatmentA", "TreatmentA", "TreatmentA", 
"TreatmentB", "TreatmentA", "TreatmentB", "TreatmentA", "TreatmentA", 
"TreatmentB", "TreatmentA", "TreatmentB"), count = c(6, 6, 6, 
6, 6, 6, 6, 2, 6, 1, 2)), row.names = c(NA, 11L), class = "data.frame")

解决方案1：使用tidyverse工具包（推荐） #

tidyr::complete() 可以直接生成指定的全量组合，并批量填充缺失值：

library(tidyverse)

# 生成cluster 1-10与两种treatment的完整组合，缺失的count填充为0
completed_df <- df %>%
  complete(cluster = 1:10, treatment = c("TreatmentA", "TreatmentB"),
           fill = list(count = 0))

解决方案2：使用Base R #

如果不想加载额外工具包，用基础R也能实现：

# 生成所有cluster和treatment的全量组合
full_combinations <- expand.grid(
  cluster = 1:10,
  treatment = c("TreatmentA", "TreatmentB"),
  stringsAsFactors = FALSE
)

# 合并原始数据，将缺失的count值替换为0
completed_df <- merge(full_combinations, df, by = c("cluster", "treatment"), all.x = TRUE)
completed_df$count[is.na(completed_df$count)] <- 0

# 按cluster排序（可选，与目标数据框顺序对齐）
completed_df <- completed_df[order(completed_df$cluster, completed_df$treatment), ]
row.names(completed_df) <- NULL

结果说明 #

处理后的数据框会包含cluster 1-10的每个值对应的TreatmentA和TreatmentB行，原始数据中不存在的组合会自动填充count = 0，与目标结构完全一致。

内容的提问来源于stack exchange，提问作者Megan Cole