DBSCAN

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88

from __future__ import annotations

import japanize_matplotlib
import matplotlib.pyplot as plt
import numpy as np
from numpy.typing import NDArray
from sklearn.cluster import DBSCAN
from sklearn.datasets import make_moons
from sklearn.preprocessing import StandardScaler

def run_dbscan_demo(
    n_samples: int = 600,
    noise: float = 0.08,
    eps: float = 0.3,
    min_samples: int = 10,
    random_state: int = 0,
) -> dict[str, int]:
    """DBSCAN で月型データをクラスタリングし、クラスタ数とノイズ点を調べる."""
    japanize_matplotlib.japanize()
    features, _ = make_moons(
        n_samples=n_samples,
        noise=noise,
        random_state=random_state,
    )
    features = StandardScaler().fit_transform(features)

    model = DBSCAN(eps=eps, min_samples=min_samples)
    labels = model.fit_predict(features)

    unique_labels = sorted(np.unique(labels))
    cluster_ids = [label for label in unique_labels if label != -1]
    noise_count = int(np.sum(labels == -1))

    core_mask = np.zeros(labels.shape[0], dtype=bool)
    if hasattr(model, "core_sample_indices_"):
        core_mask[model.core_sample_indices_] = True

    fig, ax = plt.subplots(figsize=(6.2, 5.2))
    palette = plt.cm.get_cmap("tab10", max(len(cluster_ids), 1))

    for order, cluster_id in enumerate(cluster_ids):
        mask = labels == cluster_id
        color = palette(order)
        ax.scatter(
            features[mask & core_mask, 0],
            features[mask & core_mask, 1],
            c=[color],
            s=36,
            edgecolor="white",
            linewidth=0.2,
            label=f"คลัสเตอร์ {cluster_id} (core)",
        )
        ax.scatter(
            features[mask & ~core_mask, 0],
            features[mask & ~core_mask, 1],
            c=[color],
            s=24,
            edgecolor="white",
            linewidth=0.2,
            marker="o",
            label=f"คลัสเตอร์ {cluster_id} (border)",
        )

    if noise_count:
        noise_mask = labels == -1
        ax.scatter(
            features[noise_mask, 0],
            features[noise_mask, 1],
            c="#9ca3af",
            marker="x",
            s=28,
            linewidth=0.8,
            label="Noise",
        )

    ax.set_title("DBSCAN บนข้อมูลรูปพระจันทร์")
    ax.set_xlabel("คุณลักษณะที่ 1")
    ax.set_ylabel("คุณลักษณะที่ 2")
    ax.grid(alpha=0.2)
    ax.legend(loc="upper right", fontsize=9)
    fig.tight_layout()
    plt.show()

    return {"n_clusters": len(cluster_ids), "n_noise": noise_count}

result = run_dbscan_demo()
print(f"จำนวนคลัสเตอร์ที่ค้นพบ: {result['n_clusters']}")
print(f"จำนวนจุด noise: {result['n_noise']}")