[R] Bind element of List of matrix or data.frame or list

最新推荐文章于 2020-05-14 19:26:31 发布

艳艳儿

最新推荐文章于 2020-05-14 19:26:31 发布

阅读量888

点赞数

CC 4.0 BY-SA版权

分类专栏： R 文章标签： R

本文链接：https://blog.youkuaiyun.com/COMEYAN/article/details/50486225

R 专栏收录该内容

27 篇文章

订阅专栏

本文探讨了如何使用 R 语言将矩阵或数据框列表中的元素行进行拼接的方法，包括直接使用 do.call(rbind,)、plyr 库的 ldply 函数、data.table 库的 rbindlist 函数等，并对比了它们的效率。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

Alternative solutions for list of matrix or dataframe
Alternative solutions for list of list

1. Alternative solutions for list of matrix or data.frame

If I have a list of matrix or data.frame, we can use the following ways to bind the rows of all elements.

Firstly, I generate toy data

myList1 <- list(matrix(rnorm(2*3), ncol=2),
                matrix(rnorm(2*3), ncol=2),
                matrix(rnorm(2*3), ncol=2),
                matrix(rnorm(2*3), ncol=2))
myList2 <- list(as.data.frame(matrix(rnorm(2*3), ncol=2)),
                as.data.frame(matrix(rnorm(2*3), ncol=2)),
                as.data.frame(matrix(rnorm(2*3), ncol=2)),
                as.data.frame(matrix(rnorm(2*3), ncol=2)))

Now I list the alternative solutions

# solution 1
result1.1 <- do.call(rbind, myList1)
head(result1.1)
result1.2 <- do.call(rbind, myList2)
head(result1.2)

# solution 2
## plyr: the split-apply-combine paradigm for R
library(plyr)
result1.2.1 <- ldply(myList1, rbind)
head(result1.2.1)
###Error: All inputs to rbind.fill must be data.frames
result1.2.2 <- rbind.fill(myList1)
head(result1.2.2)
result2.2.1 <- ldply(myList2, rbind)
head(result2.2.1)
result2.2.2 <- rbind.fill(myList2)
head(result2.2.2)

# solution 3
## data.table: Enhanced data.frame
library(data.table)
###Error in rbindlist(myList1) : Item 1 of list input is not a data.frame, data.table or list
result1.3 <- rbindlist(myList1)
head(result1.3)
result2.3 <- rbindlist(myList2)
head(result2.3)

From the codes, we can see that

for matrix, only do.call(rbind, ), ldply can work.
for data.frame, all do.call(rbind, ), ldply, rbind.fill, rbindlist can work.

And now benchmark for all solutions

# benchmark
## benchmark: a simple wrapper around system.time
library(rbenchmark)
benchmark(do.call(rbind, myList2), ldply(myList2, rbind), rbind.fill(myList2), rbindlist(myList2))

2. Alternative solutions for list of list

#############list of list#############################

# generate list of list
myList3 <- vector("list", 4)
for(i in 1:4){
  myList3[[i]] <- vector("list", 2)
  for(j in 1:2){
    myList3[[i]][[j]] <- rnorm(3)
  }
}

# bind each element of outer list
tempList <- lapply(myList3, function(z)do.call(rbind,z))
## selection operator "["
bind.ith.rows <- function(i) do.call(rbind, lapply(tempList, "[", i, TRUE))
nr <- nrow(tempList[[1]])
lapply(1:nr, bind.ith.rows)