K均值聚类算法 基于密度的空间聚类与噪声应用

import pandas as pd from scipy.cluster.hierarchy import linkage, dendrogram, fcluster from sklearn.preprocessing import MinMaxScaler from sklearn.metrics import silhouette_score import matplotlib.pyplot as plt data = pd.read_csv('2-game_sales_cleaned.csv') # NA使用平均值填充异常值 plt.boxplot(data['NA_Sales'], vert=False) plt.title('2-NA_Sales Distribution ') plt.show() #EU使用平均值填充异常值 plt.boxplot(data['EU_Sales'], vert=False) plt.title('2-EU_Sales Distribution ') plt.show() #JP使用平均值填充异常值 plt.boxplot(data['JP_Sales'], vert=False) plt.title('2-JP_Sales Distribution ') plt.show() # 选择用于聚类的特征列 features = ['JP_Sales', 'EU_Sales', 'NA_Sales'] X = data[features] # 因为销量值的范围是0到2，所以可以使用MinMaxScaler标准化，虽然在这个范围内标准化并不是必须的 scaler = MinMaxScaler() X_scaled = scaler.fit_transform(X) # 执行AGNES层次聚类 linked = linkage(X_scaled, 'average') # 尝试不同的聚类数量并计算轮廓系数 silhouette_avg = [] for i in range(2, 10): # 假设我们测试2到9个聚类 clusters = fcluster(linked, t=i, criterion='maxclust') score = silhouette_score(X_scaled, clusters) silhouette_avg.append(score) # 找出轮廓系数最大的聚类数 optimal_clusters = silhouette_avg.index(max(silhouette_avg)) + 2 print(f"Optimal number of clusters: {optimal_clusters}") # 使用最佳聚类数进行最终聚类 final_clusters = fcluster(linked, t=optimal_clusters, criterion='maxclust') # 使用dendrogram可视化层次聚类 plt.figure(figsize=(10, 7)) dendrogram(linked, orientation='top', labels=data.index, distance_sort='descending', show_leaf_counts=True) plt.title('Hierarchical Clustering Dendrogram') plt.xlabel('Sample index') plt.ylabel('Distance') plt.show() # 选择两个特征进行二维可视化 plt.scatter(X_scaled[:, 0], X_scaled[:, 1], c=final_clusters, cmap='viridis', alpha=0.7) plt.xlabel('Japan Sales (Scaled)') plt.ylabel('Europe Sales (Scaled)') plt.title('AGNES Clustering Results') plt.colorbar(label='Cluster') plt.show() from mpl_toolkits.mplot3d import Axes3D # 创建三维散点图 fig = plt.figure() ax = fig.add_subplot(111, projection='3d') ax.scatter(X_scaled[:, 0], X_scaled[:, 1], X_scaled[:, 2], c=final_clusters, cmap='viridis', alpha=0.7) ax.set_xlabel('Japan Sales (Scaled)') ax.set_ylabel('Europe Sales (Scaled)') ax.set_zlabel('NA Sales (Scaled)') ax.set_title('3D View of AGNES Clustering Results') plt.show()