"""
This module implements the edited nearest neighbors technique.
"""
import numpy as np
from ..base import mode, coalesce
from ._noisefilter import NoiseFilter
from ..base import NearestNeighborsWithMetricTensor
from .._logger import logger
_logger = logger
__all__ = ["EditedNearestNeighbors"]
[docs]
class EditedNearestNeighbors(NoiseFilter):
"""
References:
* BibTex::
@article{smoteNoise0,
author = {Batista, Gustavo E. A. P. A. and Prati,
Ronaldo C. and Monard, Maria Carolina},
title = {A Study of the Behavior of Several Methods for
Balancing Machine Learning Training Data},
journal = {SIGKDD Explor. Newsl.},
issue_date = {June 2004},
volume = {6},
number = {1},
month = jun,
year = {2004},
issn = {1931-0145},
pages = {20--29},
numpages = {10},
url = {http://doi.acm.org/10.1145/1007730.1007735},
doi = {10.1145/1007730.1007735},
acmid = {1007735},
publisher = {ACM},
address = {New York, NY, USA}
}
"""
[docs]
def __init__(self, remove="both", nn_params=None, n_jobs=1, **_kwargs):
"""
Constructor of the noise removal object
Args:
remove (str): class to remove from 'both'/'min'/'maj'
nn_params (dict): additional parameters for nearest neighbor calculations, any
parameter NearestNeighbors accepts, and additionally use
{'metric': 'precomputed', 'metric_learning': '<method>', ...}
with <method> in 'ITML', 'LSML' to enable the learning of
the metric to be used for neighborhood calculations
n_jobs (int): number of parallel jobs
"""
super().__init__()
self.check_isin(remove, "remove", ["both", "min", "maj"])
self.check_n_jobs(n_jobs, "n_jobs")
self.remove = remove
self.nn_params = coalesce(nn_params, {})
self.n_jobs = n_jobs
[docs]
def get_params(self, deep=False):
return {
"remove": self.remove,
"nn_params": self.nn_params,
"n_jobs": self.n_jobs,
**NoiseFilter.get_params(self, deep),
}
[docs]
def remove_noise(self, X, y):
"""
Removes noise
Args:
X (np.array): features
y (np.array): target labels
Returns:
np.array, np.array: cleaned features and target labels
"""
_logger.info("%s: Running noise removal", self.__class__.__name__)
self.class_label_statistics(y)
if len(X) < 4:
_logger.info(
"%s: Not enough samples for noise removal", self.__class__.__name__
)
return X.copy(), y.copy()
nn_params = {**self.nn_params}
nn_params["metric_tensor"] = self.metric_tensor_from_nn_params(nn_params, X, y)
nnmt = NearestNeighborsWithMetricTensor(
n_neighbors=4, n_jobs=self.n_jobs, **nn_params
)
indices = nnmt.fit(X).kneighbors(X, return_distance=False)
to_remove = []
for idx in range(len(X)):
if not y[idx] == mode(y[indices[idx][1:]]):
if (
self.remove == "both"
or (self.remove == "min" and y[idx] == self.min_label)
or (self.remove == "maj" and y[idx] == self.maj_label)
):
to_remove.append(idx)
return np.delete(X, to_remove, axis=0), np.delete(y, to_remove)