Anomaly Detector using FFT with Binary Segmentation Cutoff

This function implements an anomaly detection method that combines the Fast Fourier daltoolbox::transform (FFT) with a spectral cutoff strategy based on the Binary Segmentation (BinSeg) algorithm for changepoint detection.

The method analyzes the power spectrum of the input time series and applies the BinSeg algorithm to identify a changepoint in the spectral density, corresponding to a shift in the frequency content. Frequencies below this changepoint are considered part of the underlying trend or noise and are removed from the signal.

The modified spectrum is then transformed back into the time domain via inverse FFT, resulting in a high-pass filtered version of the series. Anomalies are identified by measuring the distance between the original and the filtered signal, highlighting unusual deviations from the dominant signal behavior.

This function is part of the HARBINGER framework and returns an object of class hanr_fft_binseg.

Usage

hanr_fft_binseg()

Value

hanr_fft_binseg object

Examples

library(daltoolbox)

#loading the example database
data(examples_anomalies)

#Using simple example
dataset <- examples_anomalies$simple
head(dataset)
#>       serie event
#> 1 1.0000000 FALSE
#> 2 0.9689124 FALSE
#> 3 0.8775826 FALSE
#> 4 0.7316889 FALSE
#> 5 0.5403023 FALSE
#> 6 0.3153224 FALSE

# setting up time series fft detector
model <- hanr_fft_binseg()

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

detection <- detect(model, dataset$serie)

# filtering detected events
print(detection[(detection$event),])
#>    idx event    type
#> 50  50  TRUE anomaly