Column standard deviation R

Question

I was wondering if there was a built-in function in R that would compute the standard deviation for columns just like colMeans computes mean for every column. It would be simple enough to write my own mini function (a compound command that invokes things like apply with sd), but I was wondering if there was already something I could use whilst also keeping my code looking clean.

...there, I would especially recommend sgibb's `colSdColMeans` as a fast implementation. — flodel, Aug 05 '13 at 01:18
for a numeric matrix, easy enough (solutions below), but how about a solution for a dataframe that only works on numeric columns? Or uses a formula argument to select columns? `colmean(~x1+x2+x3,data=d)`? — Spacedman, Aug 05 '13 at 09:52

score 70 · Answer 1 · answered Aug 04 '13 at 21:48

70

The general idea is to sweep the function across. You have many options, one is apply():

R> set.seed(42)
R> M <- matrix(rnorm(40),ncol=4)
R> apply(M, 2, sd)
[1] 0.835449 1.630584 1.156058 1.115269
R>

answered Aug 04 '13 at 21:48

Dirk Eddelbuettel

360,940
56
644
725

5

In apply(), the second argument, margin, "is a vector giving the subscripts which the function will be applied over." 2 indicates columns – rafaelvalle Nov 28 '14 at 03:11

score 16 · Answer 2 · answered Aug 05 '13 at 03:06

16

Use colSds function from matrixStats library.

library(matrixStats)
set.seed(42)
M <- matrix(rnorm(40),ncol=4)
colSds(M)

[1] 0.8354488 1.6305844 1.1560580 1.1152688

answered Aug 05 '13 at 03:06

MYaseen208

22,666
37
165
309

Can you adjust this `colSds` for sums, not only for means? – Léo Léopold Hertz 준영 Nov 11 '16 at 22:21
2

This doesn't seem to work for dataframes – seeker_after_truth Oct 26 '20 at 06:29
convert data.frame to matrix – ChuckCottrill Mar 30 '21 at 23:25

Metrics · Answer 3 · 2013-08-04T23:44:21.657

If you want to use it with groups, you can use:

library(plyr)
mydata<-mtcars
ddply(mydata,.(carb),colwise(sd))



  carb      mpg       cyl      disp       hp      drat        wt     qsec        vs        am      gear
1    1 6.001349 0.9759001  75.90037 19.78215 0.5548702 0.6214499 0.590867 0.0000000 0.5345225 0.5345225
2    2 5.472152 2.0655911 122.50499 43.96413 0.6782568 0.8269761 1.967069 0.5270463 0.5163978 0.7888106
3    3 1.053565 0.0000000   0.00000  0.00000 0.0000000 0.1835756 0.305505 0.0000000 0.0000000 0.0000000
4    4 3.911081 1.0327956 132.06337 62.94972 0.4575102 1.0536001 1.394937 0.4216370 0.4830459 0.6992059
5    6       NA        NA        NA       NA        NA        NA       NA        NA        NA        NA
6    8       NA        NA        NA       NA        NA        NA       NA        NA        NA        NA

...or without grouping: `colwise(sd)(mydata)` – flodel Aug 05 '13 at 01:04 — flodel, Aug 05 '13 at 01:04

score 2 · Answer 4 · answered Aug 05 '13 at 00:59

2

The package fBasics has a function colStdevs

 require('fBasics')
 set.seed(123)
 colStdevs(matrix(rnorm(1000, mean=10, sd=1), ncol=5))
[1] 0.9431599 0.9959210 0.9648052 1.0246366 1.0351268

answered Aug 05 '13 at 00:59

user1981275

13,002
8
72
101

Column standard deviation R

4 Answers4

Linked

Related