列表中多个数据帧的摘要统计信息

Question

If I have this list

set.seed(123)
thelist <- list(a=data.frame(x1=rnorm(10), x2=rnorm(10)),
                b=data.frame(x1=rnorm(10), x2=rnorm(10)),
                c=data.frame(x1=rnorm(10), x2=rnorm(10)))

And wanted to calculate the mean of each column within each list I could do so with the following code.

sapply(do.call("rbind",thelist),mean)

How could I calculate the standard deviation, again for each column within each list (a:c), as there is no equivalent function for sd (at least to my knowledge)?

Any suggests would be appreciated.

Answer 1

A basic R solution would be using sapply twice.

For mean only it is:

t(sapply(thelist, sapply, mean))

Resulting in

            x1        x2
a  0.074625644 0.2086220
b -0.424558873 0.3220446
c -0.008715537 0.2216860

If you want both:

my_summary <- function(x){
  c(mean = mean(x), sd = sd(x))
}

as.data.frame(lapply(thelist, sapply, my_summary))

Resulting in:

           a.x1     a.x2       b.x1      b.x2         c.x1      c.x2
mean 0.07462564 0.208622 -0.4245589 0.3220446 -0.008715537 0.2216860
sd   0.95378405 1.038073  0.9308092 0.5273024  1.082518163 0.8564451

Answer 2

First, I'd make it stackable by making the name into a column:

for (i in seq_along(thelist)) thelist[[i]]$dfname <- names(thelist)[i]

Then, stack and take means with data.table :

require(data.table)
DT <- rbindlist(thelist)

DT[,lapply(.SD,mean),by=dfname]

which gives

   dfname           x1        x2
1:      a  0.074625644 0.2086220
2:      b -0.424558873 0.3220446
3:      c -0.008715537 0.2216860

---

You might also consider the summary function, though it's clunky here:

DT[,as.list(unlist(lapply(.SD,summary))),by=dfname]
#    dfname x1.Min. x1.1st Qu. x1.Median   x1.Mean x1.3rd Qu. x1.Max. x2.Min. x2.1st Qu. x2.Median x2.Mean x2.3rd Qu. x2.Max.
# 1:      a  -1.265    -0.5318  -0.07983  0.074630    0.37800   1.715 -1.9670   -0.32690    0.3803  0.2086     0.6505  1.7870
# 2:      b  -1.687    -1.0570  -0.67700 -0.424600    0.06054   1.254 -0.3805   -0.23680    0.4902  0.3220     0.7883  0.8951
# 3:      c  -1.265    -0.6377  -0.30540 -0.008716    0.56410   2.169 -1.5490   -0.03929    0.1699  0.2217     0.5018  1.5160

---

Finally, copying my old answer , you could make your own summary-stats function:

summaryfun <- function(x) list(mean=mean(x),sd=sd(x))
DT[,as.list(unlist(lapply(.SD,summaryfun))),by=dfname]
#    dfname      x1.mean     x1.sd   x2.mean     x2.sd
# 1:      a  0.074625644 0.9537841 0.2086220 1.0380734
# 2:      b -0.424558873 0.9308092 0.3220446 0.5273024
# 3:      c -0.008715537 1.0825182 0.2216860 0.8564451

Answer 3

You can combine your data as you proposed yourself and then aggregate as follows:

thelist_named <- Map(cbind, thelist, nam = names(thelist))
thelist_binded <- do.call(rbind, thelist_named)

Aggregation Part:

my_summary <- function(x){
  c(mean = mean(x), sd = sd(x))
}
aggregate(.~nam, thelist_binded, my_summary)

Result:

  nam      x1.mean        x1.sd   x2.mean     x2.sd
1   a  0.074625644  0.953784051 0.2086220 1.0380734
2   b -0.424558873  0.930809213 0.3220446 0.5273024
3   c -0.008715537  1.082518163 0.2216860 0.8564451