distance with as.dist.obj=TRUE does not work correctly when the input data has 2 rows

[Nowosad]

In the case of an input object with 2 rows, the R dist()function returns an output of length 1 (a distance between the first and the second row). However, the distance() function with as.dist.obj=TRUE returns an empty object in these case:

library(philentropy)

m1 = matrix(c(1, 2), ncol = 1)
m2 = matrix(c(1, 2, 3), ncol = 1)

# base R
d1 = dist(m1)
d1
#>   1
#> 2 1
d2 = dist(m2)
d2
#>   1 2
#> 2 1  
#> 3 2 1

# philentropy
dp1 = distance(m1, as.dist.obj = TRUE)
#> Metric: 'euclidean'; comparing: 2 vectors.
dp1
#> dist(0)
str(dp1)
#>  'dist' num(0) 
#>  - attr(*, "Labels")= chr "euclidean"
#>  - attr(*, "Size")= int 1
#>  - attr(*, "call")= language as.dist.default(m = dist, diag = diag, upper = upper)
#>  - attr(*, "Diag")= logi FALSE
#>  - attr(*, "Upper")= logi FALSE
#>  - attr(*, "method")= chr "euclidean"
dp2 = distance(m2, as.dist.obj = TRUE)
#> Metric: 'euclidean'; comparing: 3 vectors.
dp2
#>    v1 v2
#> v2  1   
#> v3  2  1

^{Created on 2021-07-29 by the reprex package (v2.0.0)}

[Nowosad]

Possible solution is to slightly modify the distance() code at

philentropy/R/distance.R

Line 968 in c15b781

if (as.dist.obj) {

:

if (as.dist.obj) {
  if (ncols == 2) {
    dist <- stats::as.dist(matrix(c(0, dp1, dp1, 0), nrow = 2), diag = diag, upper = upper)
  } else {
    dist <- stats::as.dist(dist, diag = diag, upper = upper)
  }
  attr(dist, "method") <- method
  return(dist)
}

[HajkD]

Hi Jakub,

This is a fantastic suggestion! Thank you very much for catching this!

I now added your suggestions and please feel free to test whether it is sufficient.

library(philentropy)

m1 = matrix(c(1, 2), ncol = 1)
m2 = matrix(c(1, 2, 3), ncol = 1)

dist(m1)
#> 1
#> 2 1
distance(m1, as.dist.obj = TRUE)
#> Metric: 'euclidean'; comparing: 2 vectors.
#> 1
#> 2 1
dist(m2)
#> 1 2
#> 2 1  
#> 3 2 1
distance(m2, as.dist.obj = TRUE)
#> Metric: 'euclidean'; comparing: 3 vectors.
#>    v1 v2
#> v2  1   
#> v3  2  1

P.S. Sorry for the delay on your your pull request, I am a bit delayed, but I am working on it!

[Nowosad]

Thanks!

PS No problem - I can wait.