Most efficient way to create matrix of averages

Question

Very basic question here! On a 10 x 2 data frame, I want to calculate the average of each column and place the result in a new 1 x 2 matrix.

colmeans calculates the averages, but places the result in a 2 x 1 vector. Attempting to transpose the vector gives error message "Error in transpose(AverageX) : l must be a list."

As a workaround, I created an empty 1 x 2 matrix (filled with NAs), then used rbind to merge the Averages vector with the empty matrix, and then deleted the NAs.

It works but I bet could do the same with two lines of code rather than four. Would anyone have a better way to achieve this? Thanks very much for the help.

df <- data.frame(Price = c(1219, 1218, 1220, 1216, 1217, 1218, 1218, 1207, 1206, 1205), XXX = c( 1218, 1218, 1219, 1218, 1221,  1217 , 1217, 1216, 1219, 1216))

Average <- colMeans(df, na.rm = TRUE) #result is a vector => turn it into a 1xm matrix
AverageMatrix <- matrix(nrow = 1, ncol = 2)
AverageMatrix <- rbind(AverageMatrix, Average)
AverageMatrix<- AverageMatrix[-1, , drop = FALSE]

EDIT: @akrun creating a matrix of averages and creating an average for each column of a matrix are similar yet technically different issues, hence in my opinion this is not a duplicate. The question expresses that I already knew how to calculate an average for each column of a matrix, and that I wanted a more efficient way to store that result as a matrix, a simple issue that is not addressed in the other question.

Or `matrix(colmeans(df), 1, 2)`. Note that vectors do not have dimension in R. — lmo, Apr 19 '16 at 23:11
Thank you all very much `t(colMeans(df, na.rm = TRUE))` replicates my 4 lines of code in 1, exactly what I needed. Prefer it to the Allan T answer as it is simpler. Want to put it in as a reponse so I may mark the question as answered? — Krug, Apr 19 '16 at 23:46

Rich Scriven · Accepted Answer · 2016-04-20T01:14:26.820

3

You can get a named matrix of column means by simply transposing the result from colMeans(), because (from help(t))

When x is a vector, it is treated as a column, i.e., the result is a 1-row matrix.

t(colMeans(df, na.rm = TRUE))
#       Price    XXX
# [1,] 1214.4 1217.9

edited Apr 20 '16 at 01:14

answered Apr 20 '16 at 01:09

Rich Scriven

97,041
11
181
245

score 2 · Answer 2 · answered Apr 20 '16 at 01:50

2

With data.table, we can loop through the columns and get the mean (after converting the 'data.frame' to 'data.table' (setDT(df))

library(data.table)
setDT(df)[, lapply(.SD, mean, na.rm=TRUE)]
#   Price    XXX
#1: 1214.4 1217.9

answered Apr 20 '16 at 01:50

akrun

874,273
37
540
662

score 1 · Answer 3 · answered Apr 19 '16 at 23:11

1

I use dplyr for stuff like this. Using dplyr you can:

df <- data.frame(Price = c(1219, 1218, 1220, 1216, 1217, 1218, 1218, 1207, 1206, 1205), XXX = c( 1218, 1218, 1219, 1218, 1221,  1217 , 1217, 1216, 1219, 1216))

library(dplyr)
res <- df %>% summarize(mean(Price), mean(XXX))
res
#  mean(Price) mean(XXX)
#1      1214.4    1217.9

Edit: I forgot you wanted it as a matrix:

res <- df %>% summarize(mean(Price), mean(XXX)) %>% as.matrix

answered Apr 19 '16 at 23:11

AllanT

923
11
23

Thanks a lot AllanT. Prefer the columns to not change names, and actually am applying this on hundreds of columns, hence will be using one of the other answers. Very useful nonetheless.. – Krug Apr 19 '16 at 23:50

Most efficient way to create matrix of averages

3 Answers3