distancematrix.ostinato

Module Contents

Classes

OstinatoAnytime

Implementation of the Anytime Ostinato algorithm, which can be used to find the radius profile

Functions

find_consensus_motif(series_list, m: int) → CMResult	Finds the top-1 consensus motif and corresponding distance for the given collection of series.
find_consensus_motif_subset(series_list, m: int, k: int) → CMResult	Finds the top-1 k of n consensus motif and corresponding distance for the given collection of series.

distancematrix.ostinato.CMResult

distancematrix.ostinato.find_consensus_motif(series_list, m: int) → CMResult

Finds the top-1 consensus motif and corresponding distance for the given collection of series. The consensus motif is the subsequence (extracted from one of the series), that has a match to a subsequence from each other series within a certain distance, where that distance is minimal.

This method implements the Ostinato algorithm, described in “Matrix Profile XV: Exploiting Time Series Consensus Motifs to Find Structure in Time Series Sets” by K. Kamgar, S. Gharghabi and E. Keogh.

Parameters

• series_list – list of 1-dimensional arrays

• m – length of the consensus motif

Returns

tuple containing radius, series index and subsequence index of the consensus motif

distancematrix.ostinato.find_consensus_motif_subset(series_list, m: int, k: int) → CMResult

Finds the top-1 k of n consensus motif and corresponding distance for the given collection of series. The consensus motif is the subsequence (extracted from one of the series), that has a match to a subsequence from k other series within a certain distance, where that distance is minimal.

This method implements the k of n Ostinato algorithm, described in “Matrix Profile XV: Exploiting Time Series Consensus Motifs to Find Structure in Time Series Sets” by K. Kamgar, S. Gharghabi and E. Keogh.

Note: this algorithm has not yet been optimized for speed. (Instead, consider using the Anytime Ostinato algorithm.)

Parameters

• series_list – list of 1-dimensional arrays

• m – length of the consensus motif

Returns

tuple containing radius, series index and subsequence index of the consensus motif

class distancematrix.ostinato.OstinatoAnytime(series, m: int)

Bases: object

Implementation of the Anytime Ostinato algorithm, which can be used to find the radius profile for a collection of series. Since it is an anytime algorithm, the user can choose between more accurate results or a shorter runtime.

The radius profile contains for each subsequence the minimum distance needed to match a subsequence from all other series. Given the radius profile, the top-k minimal values correspond to the top-k consensus motifs.

This algorithm is described in “Mining Recurring Patterns in Real-Valued Time Series using the Radius Profile” by D. De Paepe and S. Van Hoecke.

calculate(self, fraction: float)

Calculates a given fraction of all distances.

Experiments show that even for low fractions, the resulting radius profile will give representative approximate results. The runtime of this method scales linear with the fraction.

Parameters

fraction – fraction of values to calculate, value in [0 .. 1]

get_radii(self, k_best: int = None)

Retrieves the radius profile for each series. If the calculation was not performed completely, the returned profiles will overestimate the real radius profile.

Parameters

k_best – If specified, calculates the radius using only the k_best best matching series (instead of all series)