hdis_mp: Discord discovery using Matrix Profile

In harbinger: A Unified Time Series Event Detection Framework

Discord discovery using Matrix Profile

Description

Discovers rare, dissimilar subsequences (discords) using Matrix Profile as implemented in the tsmp package doi:10.32614/RJ-2020-021.

Usage

hdis_mp(mode = "stamp", w, qtd)

Arguments

mode	Character. Algorithm: one of "stomp", "stamp", "simple", "mstomp", "scrimp", "valmod", "pmp".
w	Integer. Subsequence window size.
qtd	Integer. Number of discords to return (>= 3 recommended).

Value

hdis_mp object.

References

• Yeh CCM, et al. (2016). Matrix Profile I/II: All-pairs similarity joins and scalable time series motifs/discrod discovery. IEEE ICDM.

• Tavenard R, et al. tsmp: The Matrix Profile in R. The R Journal (2020). doi:10.32614/RJ-2020-021

Examples

library(daltoolbox)

# Load motif/discord example data
data(examples_motifs)

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

# Configure discord discovery via Matrix Profile
model <- hdis_mp("stamp", 4, 3)

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

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

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

harbinger documentation built on Feb. 11, 2026, 9:06 a.m.