Skip to contents

Run full architecture: denoising, anomaly detection, and clustering

arch.Rd

Applies the full analysis pipeline to a time series: denoising via seqarima(), anomaly detection via anomaly(), and clustering of anomalies via run_dbscan().

Usage

arch(ts, params)

Arguments

ts

A time series (`ts`) object.

params

A list containing configuration parameters. The following fields are supported:

  • d_max: Differencing order.

  • p_max: AR order.

  • q_max: MA order.

  • fl, fu: Lower and upper cutoff frequencies for band-pass filter.

  • nmax: Maximum number of anomalies to detect (default: 100).

  • scale: IQR multiplier for outlier threshold (default: 1.5).

  • method: Anomaly detection method; one of "iqr" or "gesd".

  • decomp: Optional time decomposition method; one of "stl", "twitter", or NULL.

  • eps: Epsilon value for DBSCAN clustering.

Value

A tibble containing:

  • time, observed: Processed time and signal values.

  • raw: Original (pre-denoising) signal.

  • anomaly: Anomaly indicator (1 = anomalous, 0 = normal).

  • cluster: DBSCAN cluster label (NA for non-anomalous points).

  • GPS: GPS time stamps.

Examples

if (FALSE) { # \dontrun{
# Generate synthetic time series
fs <- 4096
t <- seq(0, 1, by = 1 / fs)
x <- sin(2 * pi * 60 * t) + rnorm(length(t), sd = 0.5)
ts_obj <- ts(x, start = t[1], frequency = fs)

# Get default parameter set
params <- config_pipe()

# Run full analysis architecture
result <- arch(ts_obj, params)
head(result)
} # }