【異常検知】 LOF(Local Outlier Factor)による外れ値検知

外れ値検出手法の一つであるLOFに関する以下の資料を読んで試してみたいと思っていたところ、scikit-learnに例題があったのでメモします。

外れ値検出のアルゴリズム Local Outlier Factor from 翔吾 大澤

www.slideshare.net

import numpy as np
import matplotlib.pyplot as plt
from sklearn.neighbors import LocalOutlierFactor

np.random.seed(42)

# Generate train data
X = 0.3 * np.random.randn(100, 2)
# Generate some abnormal novel observations
X_outliers = np.random.uniform(low=-4, high=4, size=(20, 2))
X = np.r_[X + 2, X - 2, X_outliers]

# fit the model
clf = LocalOutlierFactor(n_neighbors=20)
y_pred = clf.fit_predict(X)
# 正常を1、異常を-1と出力するようです
ANOMALY_DATA = -1
predicted_outlier_index = np.where(y_pred == ANOMALY_DATA)
predicted_outlier = X[predicted_outlier_index]

# plot the level sets of the decision function
xx, yy = np.meshgrid(np.linspace(-5, 5, 50), np.linspace(-5, 5, 50))
Z = clf._decision_function(np.c_[xx.ravel(), yy.ravel()])
Z = Z.reshape(xx.shape)

plt.title("Local Outlier Factor (LOF)")
plt.contourf(xx, yy, Z, cmap=plt.cm.Blues_r)

a = plt.scatter(X[:200, 0], X[:200, 1], c='yellow',
                edgecolor='k', s=30, marker='o')
b = plt.scatter(X[200:, 0], X[200:, 1], c='red',
                edgecolor='k', s=30, marker='o')
c = plt.scatter(predicted_outlier[:, 0], predicted_outlier[:, 1], c='blue',
                edgecolor='k', s=10, marker='x')
plt.axis('tight')
plt.xlim((-5, 5))
plt.ylim((-5, 5))
plt.legend([a, b, c],
           ["normal observations",
            "abnormal observations",
            "observations predicted as abnormal"],
           loc="upper left", prop={'size': 12})
plt.show()