R语言：为观测日期匹配三类对照组日期的实现方法

先修正初始代码的错误（原代码中sample(dates, ...)的dates应为all_dates），并统一生成非特殊日期池：

Start <- as.Date("2016-01-01")
End <- as.Date("2017-12-31")
all_dates <- seq(from = Start, to = End, by = 1)

set.seed(1)
sam1 <- sample(all_dates, 30)
sam2 <- sample(all_dates, 5)
sam3 <- sample(all_dates, 120)

# 合并所有特殊日期，生成非特殊日期池
special_dates <- c(sam1, sam2, sam3)
non_special_dates <- all_dates[!all_dates %in% special_dates]

1. 对应数量的最近日期对照组 #

针对每个特殊日期，找到非特殊日期中距离最近的日期，去重后匹配样本数量：

get_nearest_control <- function(sample_dates, non_special_pool) {
  control_dates <- c()
  for (d in sample_dates) {
    # 计算日期距离，筛选最近的非特殊日期
    dists <- abs(as.numeric(d - non_special_pool))
    nearest <- non_special_pool[which.min(dists)]
    control_dates <- c(control_dates, nearest)
    # 移除已选日期，避免重复匹配
    non_special_pool <- non_special_pool[non_special_pool != nearest]
  }
  return(control_dates)
}

# 生成各样本的对照组
control1_sam1 <- get_nearest_control(sam1, non_special_dates)
control1_sam2 <- get_nearest_control(sam2, non_special_dates)
control1_sam3 <- get_nearest_control(sam3, non_special_dates)

2. 同期（同月/前后1个月）随机非邻近日期对照组 #

划定特殊日期的当月±1个月范围，筛选非特殊且与特殊日期间隔≥1天的日期，随机抽取对应数量：

get_periodic_nonadjacent_control <- function(sample_dates, non_special_pool, adj_gap = 1) {
  control_dates <- c()
  for (d in sample_dates) {
    # 计算同期范围（限制在总周期内）
    month_start <- as.Date(paste0(format(d, "%Y-%m"), "-01"))
    period_start <- max(month_start - months(1), Start)
    period_end <- min(month_start + months(2) - days(1), End)
    # 筛选同期内非特殊、非邻近的候选日期
    candidates <- non_special_pool[non_special_pool >= period_start & 
                                    non_special_pool <= period_end & 
                                    abs(as.numeric(d - non_special_pool)) >= adj_gap]
    # 候选集为空时，放宽至仅排除特殊日期本身
    if (length(candidates) == 0) {
      candidates <- non_special_pool[non_special_pool >= period_start & non_special_pool <= period_end]
    }
    selected <- sample(candidates, 1)
    control_dates <- c(control_dates, selected)
    non_special_pool <- non_special_pool[non_special_pool != selected]
  }
  return(control_dates)
}

# 生成各样本的对照组
control2_sam1 <- get_periodic_nonadjacent_control(sam1, non_special_dates)
control2_sam2 <- get_periodic_nonadjacent_control(sam2, non_special_dates)
control2_sam3 <- get_periodic_nonadjacent_control(sam3, non_special_dates)

3. 任意非特殊日期对照组 #

直接从非特殊日期池中随机抽取对应数量的日期：

control3_sam1 <- sample(non_special_dates, length(sam1))
control3_sam2 <- sample(non_special_dates, length(sam2))
control3_sam3 <- sample(non_special_dates, length(sam3))

内容的提问来源于stack exchange，提问作者Essi