Implementation:Scikit learn Scikit learn LocalOutlierFactor

Knowledge Sources	Scikit_learn Scikit-learn Docs
Domains	Machine Learning, Anomaly Detection
Last Updated	2026-02-08 15:00 GMT

Overview

Concrete tool for unsupervised outlier detection using the Local Outlier Factor algorithm provided by scikit-learn.

Description

LocalOutlierFactor (LOF) is an unsupervised anomaly detection method that measures the local deviation of the density of a given sample with respect to its neighbors. It identifies samples that have a substantially lower density than their neighbors as outliers. LOF computes the local density using the distance to the k-nearest neighbors, making it effective for detecting outliers in datasets with varying density clusters.

Usage

Use LocalOutlierFactor when you need to identify outliers in an unlabeled dataset based on local density deviation. It is particularly useful when outliers are defined relative to their local neighborhood rather than by global statistics. Set novelty=True to use it for novelty detection on new unseen data.

Code Reference

Source Location

• Repository: scikit-learn
• File: sklearn/neighbors/_lof.py

Signature

class LocalOutlierFactor(KNeighborsMixin, OutlierMixin, NeighborsBase):
    def __init__(
        self,
        n_neighbors=20,
        *,
        algorithm="auto",
        leaf_size=30,
        metric="minkowski",
        p=2,
        metric_params=None,
        contamination="auto",
        novelty=False,
        n_jobs=None,
    ):

Import

from sklearn.neighbors import LocalOutlierFactor

I/O Contract

Inputs

Name	Type	Required	Description
n_neighbors	int	No	Number of neighbors to use (default=20)
algorithm	str	No	Algorithm for computing nearest neighbors: 'auto', 'ball_tree', 'kd_tree', 'brute'
leaf_size	int	No	Leaf size for BallTree or KDTree (default=30)
metric	str or callable	No	Distance metric (default='minkowski')
p	float	No	Power parameter for Minkowski metric (default=2)
metric_params	dict or None	No	Additional keyword arguments for the metric function
contamination	float or str	No	Proportion of outliers in the dataset, or 'auto' (default='auto')
novelty	bool	No	Whether to use LOF for novelty detection (default=False)
n_jobs	int or None	No	Number of parallel jobs

Outputs

Name	Type	Description
negative_outlier_factor_	ndarray of shape (n_samples,)	Opposite of the LOF scores for training samples (negative values; more negative means more abnormal)
n_neighbors_	int	Actual number of neighbors used for kneighbors queries
offset_	float	Offset used to obtain binary labels from the raw LOF scores
effective_metric_	str	The effective metric used
effective_metric_params_	dict	The effective metric parameters
n_features_in_	int	Number of features seen during fit

Usage Examples

Basic Usage

from sklearn.neighbors import LocalOutlierFactor
import numpy as np

# Generate data with outliers
rng = np.random.RandomState(42)
X_inliers = rng.randn(100, 2)
X_outliers = rng.uniform(low=-4, high=4, size=(10, 2))
X = np.vstack([X_inliers, X_outliers])

# Fit and predict
lof = LocalOutlierFactor(n_neighbors=20, contamination=0.1)
y_pred = lof.fit_predict(X)
print(f"Outliers detected: {(y_pred == -1).sum()}")
print(f"LOF scores: {lof.negative_outlier_factor_[:5]}")

Related Pages

• Principle:Scikit_learn_Scikit_learn_Anomaly_Detection