I am wondering whether:

• The shape of row_ind and col_ind may be specified in get_indices for sklearn.base.BiclusterMixin.

• To remove the score method in sklearn.base.DensityMixin since is empty.

• The data type of X (np.int64 or np.float64) and y (np.int64 in classification and np.float64 in regression) can be specified in score methods, and use {array-like, sparse matrix, dataframe} instead of array-like.

• To link terms (e.g., classifier) to the glossary, since this may help to understand the base classes.

* The shape of `row_ind` and `col_ind` may be specified in `get_indices` for `sklearn.base.BiclusterMixin`.

if there are constraints in the code, then yes.

* To remove the `score` method in `sklearn.base.DensityMixin` since is empty.

changing the interface of existing public classes is tricky and we tend to avoid.

* The data type of `X` (`np.int64` or `np.float64`) and `y` (`np.int64` in classification and `np.float64` in regression) can be specified in `score` methods, and use `{array-like, sparse matrix, dataframe}` instead of `array-like`.

if the input is converted to the desired input type regardless of the input type then the input's type doesn't matter.

* To link terms (e.g., _classifier_) to the glossary, since this may help to understand the base classes.

it does help in some cases, but please be mindful of the number of hyperlinks when you do it since since it can make the doc look less readable.

Thanks for the review @adrinjalali!

I have applied the suggested changes and a few that I think are helpful.

@adrinjalali do you recall anything about this case

No I don't, but I know I myself would write "estimator instance" instead of "BaseEstimator" just because I find it more intuitive for users, but I really don't mind either of them.

+1 for "estimator instance" as well. The existence of BaseEstimator is an implementation detail for most users.

+1 of using estimator instance. I think that the user usually knows the concept of estimator (and instance), but they are not worried about implementation details (private API).

I will apply these changes!