✔️ Linting Passed

All linting checks passed. Your pull request is in excellent shape! ☀️

_{Generated for commit: b9b78ea. Link to the linter CI: here}

The change is likely from scipy/scipy#18530, I need some more time to digest it.

I am going to ping @jjerphan in case you have any insights on this, since you were involved in the COO 1d scipy PR review scipy/scipy#18530.

Here is my summary as far as scikit-learn is concerned:

• in the scikit-learn tests we are sometimes creating sparse containers (most of the time csc or csr I think) from 1d array-like. I am not sure if this is sloppy coding from us or whether it makes sense in some cases ...
• csc_matrix([1, 2, 3]) and csc_array([1, 2, 3]) both work fine in the latest scipy release (this creates a sparse container with a single row) but since ENH: sparse: Generalize coo_array to support 1d shapes scipy/scipy#18530 was merged, csc_array([1, 2, 3]) raises an exception ValueError: Cannot convert a 1d sparse array to csc format. Since some of our tests are parametrized on both sparse arrays and sparse matrices they started failing with sparse arrays. I am not sure whether the side-effect of csc_array behaving differently than csc_matrix is an unintended side-effect of the scipy COO 1d PR. It is probably doable to fix our tests to avoid this pattern.

I think we should be explicit and create sparse arrays/matrix from 2d datastructures only, unless we precisely want to test the behavior of 1d datastructures.

There aren't that many tests with this failure:

ERROR cluster/tests/test_hierarchical.py - ValueError: Cannot convert a 1d sparse array to csr format
ERROR cluster/tests/test_hierarchical.py - ValueError: Cannot convert a 1d sparse array to csr format
ERROR cluster/tests/test_hierarchical.py - ValueError: Cannot convert a 1d sparse array to csr format
ERROR cluster/tests/test_hierarchical.py - ValueError: Cannot convert a 1d sparse array to csr format
ERROR metrics/tests/test_dist_metrics.py - ValueError: Cannot convert a 1d sparse array to csr format
ERROR metrics/tests/test_dist_metrics.py - ValueError: Cannot convert a 1d sparse array to csr format
ERROR metrics/tests/test_dist_metrics.py - ValueError: Cannot convert a 1d sparse array to csr format
ERROR metrics/tests/test_dist_metrics.py - ValueError: Cannot convert a 1d sparse array to csr format
ERROR neighbors/tests/test_neighbors.py - ValueError: Cannot convert a 1d sparse array to csr format
ERROR neighbors/tests/test_neighbors.py - ValueError: Cannot convert a 1d sparse array to csr format

Merging my own PR with two approvals, there you go 😅

I added a summary of the scikit-learn tests that broke on the scipy PR in scipy/scipy#18530 (comment). I think this is best to have the discussion there with the Scipy maintainers to get their feeling about it.