• your tests are almost always only isolated points. There is no... point (!) in doing so: you're testing thousand of times the very same situation.
• inputs of the fixed tests are way too big. => 3-4 groups of 1-5 points each is way more than enough to test for proper situations. => aim for 15 points/test, NOT 50...

Hi,

the setup of your kata is currently pretty inappropriate:

• points should ideally be tuples
• if you just want to identify groups, the output should be a list (or a tuple) of sets of points (hence tuples!), and absolutely not a list of list.
• "very ideally", the output should actually even be a set of frozensets of tuples, otherwise...:
• you didn't specify the rules for the ordering of the output. Without that, valid approaches cannot pass the tests while consistent with the current requirements: [[[8, 3]], [[0, 4]], [[18, 11]], [[1, 0]]] should equal [[[1, 0]], [[8, 3]], [[0, 4]], [[18, 11]]]

if the order doesn't matter and you keep sequenced data types in the output, you'll have to sort them on your side before doing the assertions, meaning that you'll need a dedicated test method.

• you didn't say in the description if all points are unique or not in the input.

Needs random tests.

Needs sample tests.

The initial solution is for a completely different problem.