[ENH] Implementation of LOF using PyODAdapter

[aryanpola]

Reference Issues/PRs

Fixes #2109

What does this implement/fix? Explain your changes.

Implemented Local Outlier Factor using PyODadapter

PR checklist

For all contributions

• I've added myself to the list of contributors. Alternatively, you can use the @all-contributors bot to do this for you.
• The PR title starts with either [ENH], [MNT], [DOC], [BUG], [REF], [DEP] or [GOV] indicating whether the PR topic is related to enhancement, maintenance, documentation, bugs, refactoring, deprecation or governance.

[aeon-actions-bot]

Thank you for contributing to aeon

I have added the following labels to this PR based on the title: [ $\color{#FEF1BE}{\textsf{enhancement}}$ ].
I have added the following labels to this PR based on the changes made: [ $\color{#6F6E8D}{\textsf{anomaly detection}}$ ]. Feel free to change these if they do not properly represent the PR.

The Checks tab will show the status of our automated tests. You can click on individual test runs in the tab or "Details" in the panel below to see more information if there is a failure.

If our pre-commit code quality check fails, any trivial fixes will automatically be pushed to your PR unless it is a draft.

Don't hesitate to ask questions on the aeon Slack channel if you have any.

PR CI actions

These checkboxes will add labels to enable/disable CI functionality for this PR. This may not take effect immediately, and a new commit may be required to run the new configuration.

• Run pre-commit checks for all files
• Run mypy typecheck tests
• Run all pytest tests and configurations
• Run all notebook example tests
• Run numba-disabled codecov tests
• Stop automatic pre-commit fixes (always disabled for drafts)
• Disable numba cache loading
• Push an empty commit to re-run CI checks

[aryanpola]

@CodeLionX some tests ran with no window_size parameter causing errors, would it be a good idea to set the default window size if not provided, to be the same as that of stride?

[SebastianSchmidl]

Thanks for your contribution! The window_size and stride parameters should have a proper default that should get picked up during the tests. I cannot explain why this is not happening in your tests.

This already looks good. However, LOF is special because of its different calling conventions for unsupervised (outlier detection) and semi-supervised (novelty detection) problems. Please refer to my comment below on how to best integrate this into our API.

The tests in their current version are not very helpful. They mostly test the functionality of the PyODAdapter. Similar to the tests in COPOD (#2202 (comment)), I would like to test that the integration into aeon works as expected. We should also test that the metric-parameter works for our distance implementations and that the different calling conventions (unsupervised and semi-supervised) work correctly.

[aryanpola]

@CodeLionX The output values from aeon_lof and pyod_lof don't match. I've tried normalizing the series but the error persists.

FAILED aeon/anomaly_detection/tests/test_lof.py::test_lof_compare_with_pyod_direct - AssertionError: 
Not equal to tolerance rtol=1e-05, atol=0

Mismatched elements: 80 / 80 (100%)
Max absolute difference: 0.11725756
Max relative difference: 0.06374909
 x: array([1.516172, 1.018181, 1.222005, 1.838367, 1.574194, 1.062766,
       1.206536, 0.994702, 0.994427, 1.020478, 0.952964, 1.365051,
       1.122292, 0.956126, 1.028726, 1.004674, 1.378998, 0.982227,...
 y: array([1.433783, 1.000109, 1.184898, 1.728743, 1.486901, 1.06169 ,
       1.170488, 0.986849, 0.976589, 1.001037, 0.951811, 1.295437,
       1.097627, 0.952606, 1.020956, 1.000109, 1.308204, 0.981894,...

[aryanpola]

@CodeLionX Hey! I am unable to resolve these conflicts. I have tried everything, but I am stuck. Please tell me how to proceed from here.

[SebastianSchmidl]

Hey,

I cannot write directly to your branch, but resolving this should not be that difficult.

Don't you use pre-commit locally? It looks like you can push changes that do not follow our coding standards, and the changes by our pipeline are creating the issues for you. Please enable pre-commit in your dev box (pre-commit install).

Once, you have this set up, make the following changes:

--- a/aeon/anomaly_detection/__init__.py
+++ b/aeon/anomaly_detection/__init__.py
@@ -1,14 +1,15 @@
 """Time Series Anomaly Detection."""
 
 __all__ = [
-    "LOF" "DWT_MLEAD",
+    "DWT_MLEAD",
+    "IsolationForest",
     "KMeansAD",
+    "LeftSTAMPi",
+    "LOF",
     "MERLIN",
-    "STRAY",
     "PyODAdapter",
     "STOMP",
-    "LeftSTAMPi",
-    "IsolationForest",
+    "STRAY",
 ]

--- a/aeon/anomaly_detection/tests/test_lof.py
+++ b/aeon/anomaly_detection/tests/test_lof.py
@@ -235,8 +235,8 @@ def test_lof_semi_supervised():
     # Ensure shapes and types
     assert scores_aeon.shape == (20,)
     assert scores_pyod.shape == (20,)
-    assert scores_aeon.dtype == np.float_
-    assert scores_pyod.dtype == np.float_
+    assert scores_aeon.dtype == np.float64
+    assert scores_pyod.dtype == np.float64

Commit, and push again.

[MatthewMiddlehurst]

Seems to be a conflict here. Not sure if the unresolved conflict needs any action or not.

[SebastianSchmidl]

Seems to be a conflict here. Not sure if the unresolved conflict needs any action or not.

Fixed.

Were just some duplicated entries in __all__, still ready to merge from my side.