#DataUmbrella sprint

@lorentzenchr I don't really know how to add a test in sklearn/tests/test_random_projections.py. But I'll appreciate any help given to do so.

At the bottom of sklearn/tests/test_random_projections.py you add another function, the name has to start with test_. Usually you assert a know result. If you compare numpy arrays and are fine with very tiny deviations, you use assert_allclose(arr1, arr2). Have a look in the file itself for actual tests.
You can invoke the tests of a file or directory by calling it with pytest (which you have to install, e.g. pip install pytest):

pytest sklearn/tests/test_random_projection.py

or for a single test function inside the file

pytest sklearn/tests/test_random_projection.py::test_works_with_sparse_data

for complete scikit-learn tests suite just pytest inside the scikit-learn directory (beware: this takes a lot of time!).

Also, have a look at the development guide and pytest.

I'd suggest a test like

test_johnson_lindenstrauss_min_dim():
    """Test Johnson-Lindenstrauss for small eps.

    Regression test for #17111: before #19374, 32-bit systems would fail.
    """
    assert johnson_lindenstrauss_min_dim(100, eps=1e-5) == 368416070986

@lorentzenchr Thank you so much for such detailed feedback! Currently working on it now. Many thanks!

@lorentzenchr @ogrisel . Successfully added the sklearn/tests/test_random_projections.py test. Is this ok please?

Thank you for the help @lorentzenchr . Very much appreciated

@rth @cmarmo
Does this merged PR close issue #17111 ?

@reshamas Thanks for looking into it.