머신 러닝 (19) Hierarchical Clustering

import numpy as np

import pandas as pd

import matplotlib.pyplot as plt

import koreanize_matplotlib

import seaborn as sns

from sklearn.datasets import load_diabetes

from sklearn.model_selection import train_test_split, cross_val_score

from sklearn.preprocessing import StandardScaler

from sklearn.decomposition import PCA

from sklearn.cluster import KMeans, AgglomerativeClustering

from scipy.cluster.hierarchy import dendrogram, linkage

from sklearn.metrics import silhouette_score # silhouette_score 추가

import warnings

warnings.filterwarnings('ignore')

 diabetes = load_diabetes()

X = diabetes.data

y = diabetes.target

feature_names = diabetes.feature_names

df = pd.DataFrame(X, columns=feature_names)

df['target'] = y

X_features = df[feature_names].values

scaler = StandardScaler()

X_scaled = scaler.fit_transform(X_features)

k_range = range(2, 11)  # 2~10개 군집

sil_scores = []

for k in k_range:

    hc = AgglomerativeClustering(

        n_clusters=k,

        metric='euclidean',   # 거리 지표 (affinity -> metric으로 변경)

        linkage='ward'          # Ward linkage (연속형 변수에 적합)

    )

    labels = hc.fit_predict(X_scaled)

    score = silhouette_score(X_scaled, labels)

    sil_scores.append(score)

    print(f"k = {k:2d}, 실루엣 점수 = {score:.4f}")

# 최적 k 선택

best_k = k_range[int(np.argmax(sil_scores))]

best_score = max(sil_scores)

print("\n=== 실루엣 기준 최적 k 선택 결과 (계층적 군집) ===")

print(f"최적 군집 수 k = {best_k}, 실루엣 점수 = {best_score:.4f}")

best_hc = AgglomerativeClustering(

    n_clusters=best_k,

    metric='euclidean', # affinity -> metric으로 변경

    linkage='ward'

)

cluster_labels = best_hc.fit_predict(X_scaled)

df['cluster'] = cluster_labels

print("군집별 샘플 수 ")

display(df['cluster'].value_counts().sort_index())

print("군집별 타깃(질병 진행도) 평균")

display(df.groupby('cluster')['target'].mean())

pca = PCA(n_components=2, random_state=42)

X_pca = pca.fit_transform(X_scaled)

plt.figure(figsize=(6, 4))

scatter = plt.scatter(

    X_pca[:, 0],

    X_pca[:, 1],

    c=cluster_labels,

    alpha=0.7

)

plt.xlabel("PCA Component 1")

plt.ylabel("PCA Component 2")

plt.title(f"Diabetes 데이터셋 계층적 군집 (k={best_k}, linkage='ward')")

plt.colorbar(scatter, label="Cluster")

plt.grid(True)

plt.tight_layout()

plt.show()