K -均值特征化

聚类算法分析数据的空间分布。因此，k均值特征化创建了数据的压缩空间索引，可以在下一阶段将其输入模型。这是一个模型堆叠的例子。

示例 7-4。有和没有目标提示的 k-means 特征化

from scipy.spatial import Voronoi, voronoi_plot_2d
from sklearn.datasets import make_moons

training_data, training_labels = make_moons(n_samples=2000, noise=0.2)
kmf_hint = KMeansFeaturizer(k=100, target_scale=10).fit(training_data, 
                                                        training_labels)
kmf_no_hint = KMeansFeaturizer(k=100, target_scale=0).fit(training_data, 
                                                          training_labels)

def kmeans_voronoi_plot(X, y, cluster_centers, ax):
   """Plots the Voronoi diagram of the k-means clusters overlaid with the data"""
   ax.scatter(X[:, 0], X[:, 1], c=y, cmap='Set1', alpha=0.2)
   vor = Voronoi(cluster_centers)
   voronoi_plot_2d(vor, ax=ax, show_vertices=False, alpha=0.5)

图 7-6。使用目标类信息的 k-means 聚类（上图）和不使用（下图）

示例 7-5。使用 k-means 聚类特征进行分类

from sklearn.linear_model import LogisticRegression
from sklearn.svm import SVC
from sklearn.neighbors import KNeighborsClassifier
from sklearn.ensemble import RandomForestClassifier, GradientBoostingClassifier

### Generate some test data from the same distribution as training data
test_data, test_labels = make_moons(n_samples=2000, noise=0.3)

### Use the k-means featurizer to generate cluster features
training_cluster_features = kmf_hint.transform(training_data)
test_cluster_features = kmf_hint.transform(test_data)

### Form new input features with cluster features
training_with_cluster = scipy.sparse.hstack((training_data, 
                                             training_cluster_features))
test_with_cluster = scipy.sparse.hstack((test_data, test_cluster_features))

### Build the classifiers
lr_cluster = LogisticRegression(random_state=seed).fit(training_with_cluster, 
                                                       training_labels)
classifier_names = ['LR',
                    'kNN',
                    'RBF SVM',
                    'Random Forest',
                    'Boosted Trees']
classifiers = [LogisticRegression(random_state=seed),
               KNeighborsClassifier(5),
               SVC(gamma=2, C=1),
               RandomForestClassifier(max_depth=5, n_estimators=10, max_features=1),
               GradientBoostingClassifier(n_estimators=10, learning_rate=1.0, 
                                          max_depth=5)]
for model in classifiers:
    model.fit(training_data, training_labels)

### Helper function to evaluate classifier performance using ROC
def test_roc(model, data, labels):
    if hasattr(model, "decision_function"):
        predictions = model.decision_function(data)
    else:
        predictions = model.predict_proba(data)[:,1]
    fpr, tpr, _ = sklearn.metrics.roc_curve(labels, predictions)
    return fpr, tpr

### Plot results
import matplotlib.pyplot as plt
plt.figure()
fpr_cluster, tpr_cluster = test_roc(lr_cluster, test_with_cluster, test_labels)
plt.plot(fpr_cluster, tpr_cluster, 'r-', label='LR with k-means')

for i, model in enumerate(classifiers):
    fpr, tpr = test_roc(model, test_data, test_labels)
    plt.plot(fpr, tpr, label=classifier_names[i])
    
plt.plot([0, 1], [0, 1], 'k--')
plt.legend()