在R中按月份分组对各列（食品项）进行有放回抽样

核心思路 #

按月份分组提取每种食品的消费样本池，生成所有家庭×12个月的完整结构，再从对应月份的样本池中有放回抽样填充消费数据。

方法一：使用tidyverse工具链 #

library(tidyverse)

# 生成原始数据（与用户提供的结构一致）
data <- data.frame(matrix(NA, nrow=120, ncol=5))
names(data) <- c("ID", "Month", "Apple", "Maize", "Bread")
data$ID <- 1:120
data$Month <- rep(1:12, each=10)
# 注：若要让每户消费为独立随机值，将runif(12)改为runif(nrow(data))
data$Apple <- data$Month + 2 * runif(12) - 1   
data$Maize <- data$Month + 2 * runif(12) - 1
data$Bread <- data$Month + 2 * runif(12) - 1

# 1. 按月份整理每种食品的样本列表
monthly_samples <- data %>%
  group_by(Month) %>%
  summarise(
    Apple_samples = list(Apple),
    Maize_samples = list(Maize),
    Bread_samples = list(Bread),
    .groups = "drop"
  )

# 2. 构建全量ID×Month的目标结构
data2 <- expand_grid(ID = 1:120, Month = 1:12)

# 3. 关联样本池并抽样填充消费数据
data2 <- data2 %>%
  left_join(monthly_samples, by = "Month") %>%
  mutate(
    Apple = map_dbl(Apple_samples, ~sample(.x, size = 1, replace = TRUE)),
    Maize = map_dbl(Maize_samples, ~sample(.x, size = 1, replace = TRUE)),
    Bread = map_dbl(Bread_samples, ~sample(.x, size = 1, replace = TRUE))
  ) %>%
  select(ID, Month, Apple, Maize, Bread)

方法二：使用Base R #

# 生成原始数据
data <- data.frame(matrix(NA, nrow=120, ncol=5))
names(data) <- c("ID", "Month", "Apple", "Maize", "Bread")
data$ID <- 1:120
data$Month <- rep(1:12, each=10)
data$Apple <- data$Month + 2 * runif(12) - 1
data$Maize <- data$Month + 2 * runif(12) - 1
data$Bread <- data$Month + 2 * runif(12) - 1

# 按月份拆分每种食品的样本集合
apple_by_month <- split(data$Apple, data$Month)
maize_by_month <- split(data$Maize, data$Month)
bread_by_month <- split(data$Bread, data$Month)

# 构建全量ID×Month的目标结构
data2 <- expand.grid(ID = 1:120, Month = 1:12)

# 对每个月份的食品数据有放回抽样填充
data2$Apple <- sapply(data2$Month, function(m) {
  sample(apple_by_month[[as.character(m)]], size = 1, replace = TRUE)
})
data2$Maize <- sapply(data2$Month, function(m) {
  sample(maize_by_month[[as.character(m)]], size = 1, replace = TRUE)
})
data2$Bread <- sapply(data2$Month, function(m) {
  sample(bread_by_month[[as.character(m)]], size = 1, replace = TRUE)
})

# 调整列顺序与目标结构一致
data2 <- data2[, c("ID", "Month", "Apple", "Maize", "Bread")]

关键说明 #

• sample(..., replace = TRUE)实现有放回抽样，保证每次抽样可重复选取原数据中的值。
• 原数据中每个月的同食品消费值是重复的（因runif(12)仅生成12个值），若需要每户消费为独立随机值，将runif(12)替换为runif(nrow(data))即可。

内容的提问来源于stack exchange，提问作者heikeehee