【生物信息学】使用谱聚类（Spectral Clustering）算法进行聚类分析

import numpy as np
from sklearn.cluster import SpectralClustering
from sklearn.datasets import make_blobs
import matplotlib.pyplot as plt
from numpy import random
from sklearn import metrics

random.seed(1)
x, y = make_blobs(n_samples=400, centers=4, cluster_std=1.5)

plt.scatter(x[:, 0], x[:, 1], c=y, label=len(np.unique(y)))
plt.title("Initial Data Distribution")
plt.show()

gamma_best = 0
k_cluster_best = 0
CH = 0
for index, gamma in enumerate((1, 1.5, 2, 2.5, 5)):
    for index, k in enumerate((2, 3, 4, 5, 6)):
        y_pred = SpectralClustering(n_clusters=k, gamma=gamma).fit_predict(x)
        print("Calinski-Harabasz Score with gamma=", gamma, "n_clusters=", k, "score:",
              metrics.calinski_harabasz_score(x, y_pred))
        curr_CH = metrics.calinski_harabasz_score(x, y_pred)
        if (curr_CH > CH):
            gamma_best = gamma
            k_cluster_best = k
            CH = curr_CH

import numpy as np
from sklearn.cluster import SpectralClustering
from sklearn.datasets import make_blobs
import matplotlib.pyplot as plt
from numpy import random
from sklearn import metrics
 
 
SpectralClustering(affinity='rbf', coef0=1, degree=3, gamma=1.0,
                   kernel_params=None, n_clusters=4, n_init=10,
                   n_neighbors=10)
 
# scikit中的make_blobs方法常被用来生成聚类算法的测试数据，make_blobs会根据用户指定的特征数量、中心点数量、范围等来生成几类数据，这些数据可用于测试聚类算法的效果
random.seed(1)
# n_samples：样本数 n_features：int，可选（默认值= 2）centers：要生成的中心数或固定的中心位置 cluster_std: 聚类的标准偏差
x, y = make_blobs(n_samples=400, centers=4, cluster_std=1.5)
plt.scatter(x[:, 0], x[:, 1], c=y, label=len(np.unique(y)))
plt.title("Initial Data Distribution")
plt.show()
 
gamma_best = 0
k_cluster_best = 0
CH = 0
for index, gamma in enumerate((1, 1.5, 2, 2.5, 5)):
    for index, k in enumerate((2, 3, 4, 5, 6)):
        y_pred = SpectralClustering(n_clusters=k, gamma=gamma).fit_predict(x)
        # 卡林斯基哈拉巴斯得分（Calinski Harabasz score），本质是簇间距离与簇内距离的比值，整体计算过程与方差计算方式类似，也称为方差比标准，
        # 通过计算类内各点与类中心的距离平方和来度量类内的紧密度（类内距离），各个类中心点与数据集中心点距离平方和来度量数据集的分离度（类间距离），
        # 较高的 Calinski Harabasz 分数意味着更好的聚类
        print("Calinski-Harabasz Score with gamma=", gamma, "n_clusters=", k, "score:",
              metrics.calinski_harabasz_score(x, y_pred))
        curr_CH = metrics.calinski_harabasz_score(x, y_pred)
        if (curr_CH > CH):
            gamma_best = gamma
            k_cluster_best = k
            CH = curr_CH
 
print("best gamma:", gamma_best, "best cluster", k_cluster_best)
 
f = plt.figure()
sc = SpectralClustering(n_clusters=k_cluster_best, gamma=gamma_best).fit_predict(x)
plt.scatter(x[:, 0], x[:, 1], c=sc)
plt.title("n_clusters: " + str(k_cluster_best))
plt.show()