假设我有一个嵌套的向量列表。
lst1 <- list(`A`=c(a=1,b=1), `B`=c(a=1), `C`=c(b=1), `D`=c(a=1,b=1,c=1))
lst2 <- list(`A`=c(b=1), `B`=c(a=1,b=1), `C`=c(a=1,c=1), `D`=c(a=1,c=1))
lstX <- list(lst1, lst2)
如图所示,每个向量A,B,C,D
出现两次,a,b,c
出现在不同的频率中。
如何最有效地展平列表,以便a,b,c
在嵌套列表中对A,B,C,D
求和,或平均#summed
a b c
A 1 2 NA
B 2 1 NA
C 1 1 1
D 2 1 2
#averaged
a b c
A 0.5 1 NA
B 1 0.5 NA
C 0.5 0.5 0.5
D 1 0.5 1
,如下所示。真实列表有几十万个嵌套列表。
POST
答案 0 :(得分:5)
这是一个简单的基础R解决方案(将返回0
代替NA
s(不确定是否足够好)
temp <- unlist(lstX)
res <- data.frame(do.call(rbind, strsplit(names(temp), "\\.")), value = temp)
和的
xtabs(value ~ X1 + X2, res)
# X2
# X1 a b c
# A 1 2 0
# B 2 1 0
# C 1 1 1
# D 2 1 2
平均值
xtabs(value ~ X1 + X2, res) / length(lstX)
# X2
# X1 a b c
# A 0.5 1.0 0.0
# B 1.0 0.5 0.0
# C 0.5 0.5 0.5
# D 1.0 0.5 1.0
或者,更灵活的data.table
解决方案
library(data.table) #V1.9.6+
temp <- unlist(lstX)
res <- data.table(names(temp))[, tstrsplit(V1, "\\.")][, value := temp]
和的
dcast(res, V1 ~ V2, sum, value.var = "value", fill = NA)
# V1 a b c
# 1: A 1 2 NA
# 2: B 2 1 NA
# 3: C 1 1 1
# 4: D 2 1 2
平均值
dcast(res, V1 ~ V2, function(x) sum(x)/length(lstX), value.var = "value", fill = NA)
# V1 a b c
# 1: A 0.5 1.0 NA
# 2: B 1.0 0.5 NA
# 3: C 0.5 0.5 0.5
# 4: D 1.0 0.5 1.0
通常,您可以使用dcast
答案 1 :(得分:2)
我们也可以尝试
library(data.table)
DT1 <- rbindlist(lapply(do.call('c', lstX),
as.data.frame.list), fill=TRUE, idcol=TRUE)
DT1[, lapply(.SD, sum, na.rm=TRUE), .id]
# .id a b c
#1: A 1 2 0
#2: B 2 1 0
#3: C 1 1 1
#4: D 2 1 2
DT1[, lapply(.SD, function(x) sum(x, na.rm=TRUE)/.N), .id]
# .id a b c
#1: A 0.5 1.0 0.0
#2: B 1.0 0.5 0.0
#3: C 0.5 0.5 0.5
#4: D 1.0 0.5 1.0
答案 2 :(得分:1)
这不是最短的答案也不是最快的,但我们可以尝试这样的事情:
### Get all the vector names
names <- lapply(lstX, function(l) lapply(l, names))
names <- unique(unlist(names))
names
## [1] "a" "b" "c"
## Check if a name is missing, for example
setdiff(names, names(lstX[[1]][[1]]))
## [1] "c"
## Now we will check for every vectors within each list
## and fill the missing names with NA and order the results
lstX <- lapply(lstX, function(l) {
lapply(l, function(v) {
v[setdiff(names, names(v))] <- NA
v[order(names(v))] ## order by names to bind it without errors
})
})
lstX
## [[1]]
## [[1]]$A
## a b c
## 1 1 NA
## [[1]]$B
## a b c
## 1 NA NA
## [[1]]$C
## a b c
## NA 1 NA
## [[1]]$D
## a b c
## 1 1 1
## [[2]]
## [[2]]$A
## a b c
## NA 1 NA
## [[2]]$B
## a b c
## 1 1 NA
## [[2]]$C
## a b c
## 1 NA 1
## [[2]]$D
## a b c
## 1 NA 1
### Now we can bind it
matlist <- lapply(lstX, function(l) do.call(rbind, l))
matlist
## [[1]]
## a b c
## A 1 1 NA
## B 1 NA NA
## C NA 1 NA
## D 1 1 1
## [[2]]
## a b c
## A NA 1 NA
## B 1 1 NA
## C 1 NA 1
## D 1 NA 1
mysum <- apply(simplify2array(matlist), c(1, 2),
function(x) ifelse(all(is.na(x)), NA, sum(x, na.rm = TRUE)))
mysum
## a b c
## A 1 2 NA
## B 2 1 NA
## C 1 1 1
## D 2 1 2
### Average over list
mysum / length(res)
## a b c
## A 0.5 1.0 NA
## B 1.0 0.5 NA
## C 0.5 0.5 0.5
## D 1.0 0.5 1.0
修改强>
感谢@CathG,您可以像这样快速创建matlist
matlist <- lapply(lstX, function(x) {
t(sapply(x, function(y) {
y <- y[names]
names(y) <- names
y
}))
})