It's worth mentioning, for the contributors, to know that we prefer to follow some suggestions given in numpy's conventions on parameters: https://numpydoc.readthedocs.io/en/latest/format.html#docstring-standard (section 4. Parameters). In particular, we prefer to have the default values mentioned in the docstring.

Examples:

max_features : int or float, optional (default=1.0)

algorithm : {‘ball_tree’, ‘kd_tree’, ‘brute’, ‘auto’}, optional
    Algorithm used to compute the nearest neighbors:
        - ‘ball_tree’ will use BallTree (default)
        - ‘kd_tree’ will use KDTree
        - ‘brute’ will use a brute-force search.
        - ‘auto’ will attempt to decide the most appropriate algorithm based on the values passed to fit method.

The default values can be taken in most cases from the function definition or the __init__ of the classes. There are exceptions such as when the parameter is set as warn, in which case the actual default value needs to be found in the code, if not present in the documentation.

Priority is with the cases where the default value is not mentioned. I guess the ones which don't follow the format can be left alone for now. However, should we say, if the docstring of a class is touched for this PR, all parameters should follow the convention?

Uh oh!

DOC Record the default value of parameters #12111

Description

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions