ilof pietro (#1317)

* Merge branch 'ilof-pietro' of https://github.com/pietro-tanure/river into ilof-pietro

* Update ilof_notebook.ipynb

* Modify name IncrementalLOF in __init__ file of anomaly module

* Refactor code after precommit run

* Remove window_score in score_one function of Incremental LOF (since score_one already assumes that one sample is taken into account at a certain time point).

* Remove learn_many and refactor Docstring test

* Remove import pandas since unused.

* �Refactor

* Change output of `score_one` to only return one single number, not list (to comply with the implementation of other incremental anomaly detection algorithms)

* Refactor Docstring test

* Refactor from IncrementalLOF to LocalOutlierFactor

* Refactor

* Refactor X_batch to x_batch to align with PEP8.

* Verbose re-wording.

* Refactor X_score to x_scores

* Remove type of returned output in DocString

* Make check_equal static method and refactor related variable names according to PEP8

* Refactor capital X variable to x_list to represent the stored list of observations.

* Refactor code with black

* Remove description for unnecessary variables.

* Re-wording for variable description.

* Refactor.

* Add learn_many to learn multiple instances at a time and update Docstring test to take into account the newly implemented function.

* Remove description of unnecessary variables.

* Spelling correction.

* Modify variable names by removing capital letters to align with PEP8.

* Refactor calc_local_reach_dist and cal_lof and change these functions to static methods.

* Change expand_objects to static method

* Refactor.

* Change define_sets to static method.

* Change calc_reach_dist_newpoints to static method.

* Change calc_reach_dist_otherpoints to static method and refactor.

* Refactor.

* Refactor calc_reach_dist_new_points

* Refactor calc_reach_dist_other_points.

* Remove import

* Refactor and add description in docstring regarding expected performance of ILOF.

* Refactor.

* Remove one unnecessary variable of calc_reach_dist_other_points.

* Modify docstring description of the algorithm

* Add comments to justify the returning results of score_one function.

* Add tests for the newly implemented iLOF algorithm.

* Remove iLOF notebook (since the content has been covered by the test file within the same module).

---------

Co-authored-by: Hoang-Anh Ngo <[email protected]>

river/anomaly/__init__.py

@@ -17,6 +17,7 @@
 from .filter import QuantileFilter, ThresholdFilter
 from .gaussian import GaussianScorer
 from .hst import HalfSpaceTrees
+from .ilof import LocalOutlierFactor
 from .svm import OneClassSVM
 
 __all__ = [
@@ -27,4 +28,5 @@
     "OneClassSVM",
     "QuantileFilter",
     "ThresholdFilter",
+    "LocalOutlierFactor",
 ]