Add pairwise #1
Add pairwise #1
Conversation
|
|
||
| # pairwise(f, x[, y]) | ||
| # | ||
| # Return a matrix holding the result of applying `f` to all possible pairs |
There was a problem hiding this comment.
are we sure it will always be a matrix? Maybe it should be a more general n-dimensional pairwise (but admittedly then it should probably be tuple-wise)
There was a problem hiding this comment.
In particular as x and y are allowed to be any iterators what should happen with a 0-dimensional iterator (e.g. when x is a number)
There was a problem hiding this comment.
are we sure it will always be a matrix? Maybe it should be a more general
n-dimensional pairwise (but admittedly then it should probably be tuple-wise)
"pairwise" seems to imply there are only two elements, right? IIRC we mentioned the possibility of having a more general outer function, but I find pairwise nicer to compute pairwise correlations, and it's already used by Distances.
In particular as
xandyare allowed to be any iterators what should happen with a 0-dimensional iterator (e.g. whenxis a number)
Well the number of dimensions of the iterator doesn't matter, only the number of elements it contains. So a number gives a 1×1 matrix.
2 participants
To allow StatsBase to define the generic version (JuliaStats/StatsBase.jl#627) while Distances can continue defining a specific one for its types.
I didn't add real docstrings as the one in StatsBase is relatively complex and it would be absurd to have to update it here if the function evolves.