hanr_remd: Anomaly detector using REMD

In harbinger: A Unified Time Series Event Detection Framework

Anomaly detector using REMD

Description

Anomaly detection using REMD with EMD-based decomposition. The detector decomposes the series, selects components according to curvature, and flags large residual deviations as anomalies. Wraps the EMD-based model presented in the forecast package. The internal ARIMA adjustment is fitted on ts_data(..., sw = 1), which is the aligned single-series representation expected by raw-series forecasters in tspredit.

Usage

hanr_remd(noise = 0.1, trials = 5)

Arguments

noise	Noise amplitude for the decomposition.
trials	Number of trials used by the decomposition step.

Value

hanr_remd object

References

• Souza, J., Paixão, E., Fraga, F., Baroni, L., Alves, R. F. S., Belloze, K., Dos Santos, J., Bezerra, E., Porto, F., Ogasawara, E. REMD: A Novel Hybrid Anomaly Detection Method Based on EMD and ARIMA. Proceedings of the International Joint Conference on Neural Networks, 2024. doi:10.1109/IJCNN60899.2024.10651192

Examples

library(daltoolbox)

# Load anomaly example data
data(examples_anomalies)

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

# Configure REMD detector
model <- hanr_remd()

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

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

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

harbinger documentation built on July 10, 2026, 5:07 p.m.