来自 7 列 80 行的随机样本

Question

我有一个 7 列 80 行的表格，看起来像这样，

         **`1`   `3`   `5`   `7`   `9`  `11`
          <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>**
      1     6     7     7     8    NA    NA
      2     6     7    13    13    14    NA
      3     9    10    10     8    NA    NA
      4     4     3     5     3    NA    NA
      5     3     2     5     5     6    NA
      6     7     7     5     4     5     5
      7     9     5     8     8    NA    NA
      8     6     6     7    NA     7    NA
      9    NA     6     5     5    NA    NA
     10     6     7     6     4     7     6
      # ... with 70 more rows

我需要对每行的 7 列之一进行随机抽样。例如第 1 (8) 行、第 2 (6) 行、第 3 (10) 行等所有 80 行。我可以使用示例功能吗？如果可以，那么如何使用？我可以用 NA 做什么。我需要进行 1000 次采样并计算每个样本的平均值。

任何帮助将不胜感激！ 谢谢， 阿尔丁

Answer 1

我们可以使用apply遍历行，获取非NA元素并获取sample

n <- 1000
lst <- replicate(n, apply(df1, 1, function(x) sample(x[!is.na(x)], 1)),
               simplify = FALSE)
Reduce(`+`, lst)/n

---

或者pmap和rowMeans

library(tidyverse)
rowMeans(replicate(n, pmap_int(df1, ~
                          c(...) %>% 
                          na.omit %>%
                          sample(., 1))))

数据

set.seed(24)
df1 <- as.data.frame(matrix(sample(c(1:9, NA), 80 * 7, replace = TRUE), 80, 7))

Answer 2

使用sapply()。

sapply(as.data.frame(t(df1)), function(x) sample(na.omit(x), 1))

数据

df1 <- structure(list(X.1. = c(6L, 6L, 9L, 4L, 3L, 7L, 9L, 6L, NA, 6L
), X.3. = c(7L, 7L, 10L, 3L, 2L, 7L, 5L, 6L, 6L, 7L), X.5. = c(7L, 
13L, 10L, 5L, 5L, 5L, 8L, 7L, 5L, 6L), X.7. = c(8L, 13L, 8L, 
3L, 5L, 4L, 8L, NA, 5L, 4L), X.9. = c(NA, 14L, NA, NA, 6L, 5L, 
NA, 7L, NA, 7L), X.11. = c(NA, NA, NA, NA, NA, 5L, NA, NA, NA, 
6L)), class = "data.frame", row.names = c("1", "2", "3", "4", 
"5", "6", "7", "8", "9", "10"))

Answer 3

这是一个使用 plyr::adply 的解决方案。

library(plyr)

# original dataset
df1 <- data.frame(
   c( 6,  6,  9,  4,  3,  7,  9,  6, NA, 6),
   c( 7,  7, 10,  3,  2,  7,  5,  6,  6, 7),
   c( 7, 13, 10,  5,  5,  5,  8,  7,  5, 6),
   c( 8, 13,  8,  3,  5,  4,  8, NA,  5, 4),
   c(NA, 14, NA, NA,  6,  5, NA,  7, NA, 7),
   c(NA, NA, NA, NA, NA,  5, NA, NA, NA, 6)
)


# returns a single column from a row with NA's removed
samplerow <- function(r) {
  # r is a single row of df
  # eliminate NAs from the dataset.
  r <- r[!is.na(r)]
  # Return one sample from this row
  # Not sure what happens if the row is all NAs. Don't do that.
  r[sample.int(length(r),1)]
}

N <- 1000
# for N times,
# for each row select 1 non-NA valued column,
# take the mean of all rows
replicate(N, mean(adply(df1, 1, samplerow, .expand=F)$V1))
#...redacted...
N <- 5
set.seed(1)
replicate(N, mean(adply(df1, 1, samplerow, .expand=F)$V1))
[1] 6.0 6.2 6.2 7.0 7.1