hanct_kmeans: Anomaly detector using k-means

In harbinger: A Unified Time Series Event Detection Framework

Anomaly detector using k-means

Description

Distance-based anomaly and discord detection using k-means clustering. The detector clusters the series and flags observations or subsequences that are far from the nearest centroid. When seq equals one, isolated observations are labeled as anomalies. When seq is greater than one, subsequences are labeled as discords. Wraps the kmeans implementation from the stats package.

Usage

hanct_kmeans(seq = 1, centers = NA)

Arguments

seq	sequence size
centers	number of centroids

Value

hanct_kmeans object

References

• Ogasawara, E., Salles, R., Porto, F., Pacitti, E. Event Detection in Time Series. 1st ed. Cham: Springer Nature Switzerland, 2025. doi:10.1007/978-3-031-75941-3

Examples

library(daltoolbox)

# Load anomaly example data
data(examples_anomalies)

# Use a simple example
dataset <- examples_anomalies$simple
head(dataset)

# Configure k-means detector
model <- hanct_kmeans()

# Fit the model
model <- fit(model, dataset$serie)

# Run detection
detection <- detect(model, dataset$serie)

# Show detected events
print(detection[(detection$event),])

harbinger documentation built on May 14, 2026, 5:06 p.m.