将函数应用于循环中的成对的列

Question

My data look like this:

a1 <- runif(30, 1, 100)
b1 <- runif(30, 1, 100)
c1 <- runif(30, 1, 100)
a2 <- runif(30, 1, 100)
b2 <- runif(30, 1, 100)
c2 <- runif(30, 1, 100)
dframe <- data.frame(a1=a1, b1=b1, c1=c1, a2=a2, b2=b2, c2=c2)

I want to calculate the correlation between a1 and a2, b1 and b2, c1 and c2, but I'd like to do it in an efficient way, avoiding writing one line of code for each correlation. I tried to write a for loop but I did not succeed.

Answer 1

Here is an option

lapply(split.default(dframe, sub("\\d+$", "", names(dframe))), cor)
#$a
#          a1        a2
#a1 1.0000000 0.1132033
#a2 0.1132033 1.0000000

#$b
#           b1         b2
#b1 1.00000000 0.09113974
#b2 0.09113974 1.00000000

#$c
#           c1         c2
#c1  1.0000000 -0.2066311
#c2 -0.2066311  1.0000000

We split your data frame column-wise and then iterate over the resulting list with lapply .

Answer 2

A base R idea,

sapply(unique(gsub('\\d+', '', names(dframe))), function(i) 
                                                cor(dframe[grepl(i, names(dframe))]))

which gives,

  abc [1,] 1.00000000 1.0000000 1.00000000 [2,] 0.01987806 -0.2247265 -0.08667891 [3,] 0.01987806 -0.2247265 -0.08667891 [4,] 1.00000000 1.0000000 1.00000000 

Answer 3

In a tidyverse style,

set.seed(123)
# for a reproducible way, set seeds.
a1 <- runif(30, 1, 100)
b1 <- runif(30, 1, 100)
c1 <- runif(30, 1, 100)
a2 <- runif(30, 1, 100)
b2 <- runif(30, 1, 100)
c2 <- runif(30, 1, 100)
dframe <- data.frame(a1=a1, b1=b1, c1=c1, a2=a2, b2=b2, c2=c2)
library(psych)
library(tidyverse)
dframe %>% 
    corr.test(use = "pairwise.complete.obs") %>% 
    .$ci %>% 
    rownames_to_column('pairs') %>% 
    filter(pairs %in% c('a1-a2','b1-b2','c1-c2'))
#>   pairs      lower            r     upper         p
#> 1 a1-a2 -0.2365720  0.135222126 0.4724741 0.4761839
#> 2 b1-b2 -0.5137963 -0.188401038 0.1843832 0.3187486
#> 3 c1-c2 -0.3523592  0.009060141 0.3681278 0.9621014

^{Created on 2018-11-08 by the reprex package (v0.2.1)}