Update olivetti_faces.py #2900
Update olivetti_faces.py #2900
Conversation
| Returns | ||
| -------- | ||
| dataset: A dot-accessible dictionary | ||
| Contains faces: array of faces in the original database. |
There was a problem hiding this comment.
the fields are actually data, images, target, DESCR as far as I can see, not faces...?
There was a problem hiding this comment.
This documentation gets processed (more-or-less) as ReStructuredText. Run make doc-noplots and look at the output in doc/_build/html/stable/modules/generated/. Using one of the RST list formats might be cleaner.
There was a problem hiding this comment.
okay that was a mistake, will change that but i don't understand one thing.
why are we sending the "faces" twice.Once in the form of column matrix "data"(if im not wrong) and then in the form of "images" ?
isn't that a waste of space?
There was a problem hiding this comment.
This is just for convenience. There is no memory copy, this is just a reshaping of the same memory buffer.
There was a problem hiding this comment.
The data attribute generally has a 2D shape (n_samples, n_features) in other load, so we respect that convention here while allowing the user to access the original shape of the images in the faces attribute.
|
This has been solved in some pr. |
i have written a little documentation for the return object in olivetti_faces.py
and changed the ordering of the parameters