循环函数避免在分层抽样中使用R

Question

我正在使用下面的代码对人口进行分层随机抽样，目标是根据性别和地区预先确定的。可以看出，相同的代码重复了三次（每次都使用了 rbind）。有没有一种方法可以编写一个循环函数来重复这个过程，这样我就不必写几次了？

library(tidyverse)
population<- data.frame(region=rep(c("North", "South", "North", "South"),50000),
                        sex=rep(c("Male", "Female", "Female", "Male"),50000))

targets<- data.frame(region=c("North", "South", "North", "South"),
                     sex=c("Male", "Female", "Female", "Male"),
                     total=c(20,25,20,30)) 


n=0
n=n+1
sample<- population %>% filter(region==targets$region[[1]], 
                           sex==targets$sex[[1]]) %>% 
  slice_sample(n=targets$total[[1]])
n=n+1
sample<- sample %>%  rbind(population %>% filter(region==targets$region[[n]], 
                               sex==targets$sex[[n]]) %>% 
  slice_sample(n=targets$total[[n]]))
n=n+1
sample<- sample %>%  rbind(population %>% filter(region==targets$region[[n]], 
                               sex==targets$sex[[n]]) %>% 
  slice_sample(n=targets$total[[n]]))
n=n+1
sample<- sample %>%  rbind(population %>% filter(region==targets$region[[n]], 
                               sex==targets$sex[[n]]) %>% 
  slice_sample(n=targets$total[[n]]))

Answer 1

感谢Harry Smith提供的帮助，我找到了答案。它包括对他的代码的一个小调整：

n = 4  
sample_list <- list()
for(i in 1:n){
  sample<- population %>% filter(region==targets$region[[i]], 
                                                   sex==targets$sex[[i]]) %>% 
                               slice_sample(n=targets$total[[i]])
  sample_list[[i]] <- sample
}

a<-bind_rows(sample_list)

Answer 2

像这样。它将 dfs 保存在列表中。

n = 4  
sample_list <- list()
for(i in 1:n){
sample<-rbind(population %>% filter(region==targets$region[[i]], 
                                                 sex==targets$sex[[i]]) %>% 
                             slice_sample(n=targets$total[[i]]))
sample_list[[i]] <- sample
}

Answer 3

假设数据中的变量为h。然后迭代循环，根据指示变量提取相应的观察值。

data <- csv.()
sizes <- c(a, b, c)
indicator <- data$h
total_size <- sum(sizes)
ratio <- sizes / total_size

创建一个空矩阵来存储最终样本。这将启动一个循环遍历每个 (i) 。根据其比率计算当前切片所需的样本大小

sample1 <- NULL
for (i in 1:length(sizes)) {
  slice <- data[ind == i,]
  slice_sample_size <- round(ratio[i] * total_size)
  slice_sample <- slice[sample(nrow(slice), slice_sample_size),]
  sample1 <- rbind(sample1, slice_sample)
}

或使用 df

slice_data <- df[df$h == slice[i], ]
for (i in 1:n_slice) {
  slice_data <- df[df$g1 == strata[i], ]
  slice_sample <- slice_data[sample.int(nrow(slice_data), sample_sizes[i]), ]