التجميع: K-Means وDBSCAN

from sklearn.cluster import KMeans, DBSCAN, AgglomerativeClustering
from sklearn.metrics import silhouette_score
from sklearn.preprocessing import StandardScaler
from sklearn.datasets import make_blobs
import numpy as np

class="tok-comment"># ── Sample data ────────────────────────────────────────────────────────
X, _ = make_blobs(n_samples=class="tok-num">300, centers=class="tok-num">4, cluster_std=class="tok-num">0.8, random_state=class="tok-num">42)

class="tok-comment"># ── K-Means with K selection via silhouette ────────────────────────
X_scaled = StandardScaler().fit_transform(X)

best_k, best_score = class="tok-num">2, -class="tok-num">1
for k in range(class="tok-num">2, class="tok-num">11):
    km = KMeans(n_clusters=k, init=class="tok-str">'k-means++', n_init=class="tok-num">10, random_state=class="tok-num">42)
    labels = km.fit_predict(X_scaled)
    score = silhouette_score(X_scaled, labels)
    if score > best_score:
        best_k, best_score = k, score

km_final = KMeans(n_clusters=best_k, init=class="tok-str">'k-means++', n_init=class="tok-num">10, random_state=class="tok-num">42)
labels_km = km_final.fit_predict(X_scaled)
print(fclass="tok-str">"K={best_k}, silhouette={best_score:.3f}, inertia={km_final.inertia_:.1f}")

class="tok-comment"># ── DBSCAN — auto-detects clusters and outliers ────────────────────
db = DBSCAN(eps=class="tok-num">0.5, min_samples=class="tok-num">5)
labels_db = db.fit_predict(X_scaled)
n_clusters = len(set(labels_db)) - (class="tok-num">1 if -class="tok-num">1 in labels_db else class="tok-num">0)
n_noise    = list(labels_db).count(-class="tok-num">1)
print(fclass="tok-str">"DBSCAN: {n_clusters} clusters, {n_noise} noise points")

class="tok-comment"># ── Agglomerative — no K needed upfront (use dendrogram) ──────────
from scipy.cluster.hierarchy import dendrogram, linkage
Z = linkage(X_scaled, method=class="tok-str">'ward')
class="tok-comment"># Cut at K=class="tok-num">3
agg = AgglomerativeClustering(n_clusters=class="tok-num">3, linkage=class="tok-str">'ward')
labels_agg = agg.fit_predict(X_scaled)