Implementation:Scikit learn Scikit learn OPTICS

Knowledge Sources	Scikit_learn Scikit-learn Docs
Domains	Clustering, Density-Based Clustering
Last Updated	2026-02-08 15:00 GMT

Overview

Concrete tool for performing ordering points to identify clustering structure (OPTICS) provided by scikit-learn.

Description

OPTICS is a density-based clustering algorithm closely related to DBSCAN that finds core samples of high density and expands clusters from them. Unlike DBSCAN, OPTICS keeps the cluster hierarchy for a variable neighborhood radius, making it better suited for datasets with clusters of varying densities. Clusters are extracted from the computed cluster-order using either a DBSCAN-like method or the automated xi technique. The implementation first performs k-nearest-neighbor searches on all points, then computes reachability distances to construct the cluster order.

Usage

Use OPTICS when you expect clusters of varying density, when you want to explore the hierarchical density structure of data, or when DBSCAN fails because a single eps value cannot capture clusters at different density scales. It is well-suited for large datasets and spatial data analysis.

Code Reference

Source Location

• Repository: scikit-learn
• File: sklearn/cluster/_optics.py

Signature

class OPTICS(ClusterMixin, BaseEstimator):
    def __init__(
        self,
        *,
        min_samples=5,
        max_eps=np.inf,
        metric="minkowski",
        p=2,
        metric_params=None,
        cluster_method="xi",
        eps=None,
        xi=0.05,
        predecessor_correction=True,
        min_cluster_size=None,
        algorithm="auto",
        leaf_size=30,
        memory=None,
        n_jobs=None,
    ):

Import

from sklearn.cluster import OPTICS

I/O Contract

Inputs

Name	Type	Required	Description
min_samples	int or float	No	Number of samples in a neighborhood for a core point. Default is 5.
max_eps	float	No	Maximum distance between two samples for neighborhood. Default is np.inf.
metric	str or callable	No	Distance metric. Default is "minkowski".
p	int	No	Power parameter for the Minkowski metric. Default is 2.
metric_params	dict or None	No	Additional keyword arguments for the metric function. Default is None.
cluster_method	str	No	Extraction method: "xi" or "dbscan". Default is "xi".
eps	float or None	No	Maximum distance for DBSCAN cluster extraction method. Default is None.
xi	float	No	Steepness threshold for xi cluster extraction (between 0 and 1). Default is 0.05.
predecessor_correction	bool	No	Correct clusters based on predecessor info. Default is True.
min_cluster_size	int or float or None	No	Minimum number of samples in a cluster. Default is None (uses min_samples).
algorithm	str	No	Nearest neighbor algorithm: "auto", "ball_tree", "kd_tree", or "brute". Default is "auto".
leaf_size	int	No	Leaf size for BallTree or KDTree. Default is 30.
memory	str or joblib.Memory	No	Used to cache nearest neighbor computations. Default is None.
n_jobs	int or None	No	Number of parallel jobs. Default is None.

Outputs

Name	Type	Description
labels_	ndarray of shape (n_samples,)	Cluster labels. Noisy samples and unclustered points are labeled -1.
reachability_	ndarray of shape (n_samples,)	Reachability distances per sample, indexed by ordering_.
ordering_	ndarray of shape (n_samples,)	The cluster-ordered list of sample indices.
core_distances_	ndarray of shape (n_samples,)	Distance to the min_samples-th nearest neighbor for each sample.
predecessor_	ndarray of shape (n_samples,)	Point that a sample was reached from, indexed by ordering_.
cluster_hierarchy_	ndarray of shape (n_clusters, 2)	List of clusters with start and end indices in the reachability plot (xi method only).

Usage Examples

Basic Usage

from sklearn.cluster import OPTICS
import numpy as np

X = np.array([[1, 2], [2, 5], [3, 6],
              [8, 7], [8, 8], [7, 3]])

clustering = OPTICS(min_samples=2).fit(X)
print(clustering.labels_)
print(clustering.reachability_)

Related Pages

• Principle:Scikit_learn_Scikit_learn_Clustering