Time Series Anomaly Detection

Duration: 5 min

This module delves into the critical area of time series anomaly detection, a crucial aspect of data analysis that helps identify unusual patterns or outliers in time-dependent data. Understanding and detecting anomalies is essential for various applications, including fraud detection, equipment maintenance, and monitoring system health.

Understanding Time Series Anomalies

Time series anomalies are deviations from the expected behavior of a time series. These anomalies can be caused by various factors, such as sudden changes in the environment, equipment malfunctions, or unusual user behavior. Detecting these anomalies is vital for maintaining system integrity and performance.

import numpy as np
import matplotlib.pyplot as plt

# Generate synthetic time series data
np.random.seed(0)
data = np.sin(np.linspace(0, 3 * np.pi, 100)) + np.random.normal(scale=0.1, size=100)

# Introduce an anomaly
data[50] += 2

# Plot the time series
plt.plot(data)
plt.title('Time Series with Anomaly')
plt.show()

Try it in Google Colab:

A plot showing a sinusoidal time series with an anomaly at index 50.

Implementing Anomaly Detection with Isolation Forest

Isolation Forest is a machine learning algorithm that can be used for anomaly detection. It works by isolating observations by randomly selecting a feature and then randomly selecting a split value between the maximum and minimum values of the selected feature. This method is effective for identifying anomalies in time series data.

from sklearn.ensemble import IsolationForest
import numpy as np
import matplotlib.pyplot as plt

# Generate synthetic time series data
np.random.seed(0)
data = np.sin(np.linspace(0, 3 * np.pi, 100)) + np.random.normal(scale=0.1, size=100)

# Introduce an anomaly
data[50] += 2

# Reshape data for the model
data_reshaped = data.reshape(-1, 1)

# Fit the Isolation Forest model
model = IsolationForest(contamination=0.1)
model.fit(data_reshaped)

# Predict anomalies
anomaly_scores = model.decision_function(data_reshaped)
anomaly_predictions = model.predict(data_reshaped)

# Plot the time series with anomalies
plt.plot(data, label='Time Series')
plt.plot(np.arange(len(data))[anomaly_predictions == -1], data[anomaly_predictions == -1], 'ro', label='Anomalies')
plt.title('Time Series with Detected Anomalies')
plt.legend()
plt.show()

💡 Tip: When using Isolation Forest for anomaly detection, it's important to tune the contamination parameter to match the expected proportion of anomalies in your data.

Key Concepts

Check Your Understanding