CODEWITHSUNDEEP
rapplylapply
Enzo
Enzo

Reputation: 2611

Why there is a different coercion behaviour between these two use of apply & lapply

I needed to create a list out of a data.frame in a row-by-row shape for later conversion to JSON.

I was surprised to see a different behaviour between apply and lapply:

This is an example:

df_ = data.frame( Month = month.abb[1:5],
A=rnorm(5, mean = 5, sd = 4),
B=rnorm(5, mean = 5, sd = 4),
C=rnorm(5, mean = 5, sd = 4),
D=rnorm(5, mean = 5, sd = 4),
E=rnorm(5, mean = 5, sd = 4), stringsAsFactors=F)


df1_ <- lapply(1:nrow(df_), function(i) {l_ <- as.list(df_[i, ])
    return(l_)})

df2_ <- apply(df_,1, as.list)

df1_[[1]]
$Month
[1] "Jan"

$A
[1] 8.235366

$B
[1] -0.2163872

$C
[1] 10.75159

$D
[1] 5.401502

$E
[1] 7.175394

df2_[[1]]
$Month
[1] "Jan"

$A
[1] "8.2353656"

$B
[1] "-0.2163872"

$C
[1] "10.751592"

$D
[1] " 5.4015020"

$E
[1] " 7.175394"

As far as I can tell they appear identical, but apply has coerced all numeric to character, while lapply has preserved them as numeric.

I would like to have an understanding of why this happens, to avoid this pitfall in the future.

Upvotes: 0

Views: 32

Answers (1)

shadow
shadow

Reputation: 22293

apply takes matrix arguments. data.frames will be coerced to matrix before anything else is done. Hence the conversion of everything to the same type (character).

lapply takes list arguments. Therefore it coerces the data.frame to a list and does not have to convert the arguments.

Upvotes: 1

Related Questions