Signature

model = sp.DBSCAN(eps: float = 0.5, min_samples: int = 5)

model.fit(x: list[float], y: list[float]) -> None
model.fit_predict(x: list[float], y: list[float]) -> list[int]

model.labels_      -> list[int]
model.n_clusters_  -> int
model.n_noise_     -> int

Description

Low-level DBSCAN class for programmatic access to cluster labels. -1 labels indicate noise points (not part of any cluster).

Constructor Parameters

Methods

fit(x, y)

Runs DBSCAN on the 2D data. Populates labels_, n_clusters_, and n_noise_.

fit_predict(x, y) -> list[int]

Equivalent to calling fit(x, y) then returning labels_.

Attributes

Examples

Accessing labels

from seraplot import DBSCAN
import seraplot as sp
x = [1.0, 1.1, 1.2, 10.0, 10.1, 99.0]
y = [1.0, 0.9, 1.1, 10.2, 10.0, 99.0]
xy = [[xi, yi] for xi, yi in zip(x, y)]
model = DBSCAN(eps=0.5, min_samples=2)
labels = model.fit_predict(xy)
chart = sp.build_scatter_chart(
    f"DBSCAN ({model.n_clusters_} clusters)",
    x_values=x,
    y_values=y,
    color_groups=[str(lbl) for lbl in labels],
)

from seraplot import DBSCAN
const sp = require('seraplot');
const x = [1.0, 1.1, 1.2, 10.0, 10.1, 99.0]
const y = [1.0, 0.9, 1.1, 10.2, 10.0, 99.0]
const xy = [[xi, yi] for xi, yi in zip(x, y)]
const model = DBSCAN({eps: 0.5, min_samples: 2})
const labels = model.fit_predict(xy)
const chart = sp.build_scatter_chart(f"DBSCAN ({model.n_clusters_} clusters)",
x,
{
    y_values: y,
    color_groups: [str(lbl) for lbl in labels]
})

from seraplot import DBSCAN
import * as sp from 'seraplot';
const x: number[] = [1.0, 1.1, 1.2, 10.0, 10.1, 99.0]
const y: number[] = [1.0, 0.9, 1.1, 10.2, 10.0, 99.0]
const xy: number[] = [[xi, yi] for xi, yi in zip(x, y)]
const model = DBSCAN({eps: 0.5, min_samples: 2})
const labels = model.fit_predict(xy)
const chart = sp.build_scatter_chart(f"DBSCAN ({model.n_clusters_} clusters)",
x,
{
    y_values: y,
    color_groups: [str(lbl) for lbl in labels]
})

Pipeline: cluster then visualize

import seraplot as sp

model = sp.DBSCAN(eps=1.0, min_samples=5)
model.fit(x_data, y_data)

color_groups = [str(lbl) for lbl in model.labels_]

chart = sp.build_scatter_chart(
    f"DBSCAN ({model.n_clusters_} clusters)",
    x_values=x_data,
    y_values=y_data,
    color_groups=color_groups,
)

Algorithmic Functioning

The DBSCAN class exposes the same Rust-backed algorithm as the chart variant.

For a point $p$, its $\epsilon$-neighbourhood is:

• Core point: $|N_\epsilon(p)| \geq \text{min_samples}$
• Border point: reachable from a core point but not itself a core point
• Noise point: not reachable from any core point — label $-1$

SeraPlot builds a KD-tree for $O(\log n)$ radius queries and expands clusters via parallel BFS with SIMD distance acceleration. n_clusters_ counts only true clusters; noise points are excluded.

See also

• DBSCAN Chart
• DBSCAN 3D
• Scatter Chart