distancematrix.valmod

Module Contents

Functions

find_variable_length_motifs(series, min_motif_length, max_motif_length, cache_size=3, noise_std=0.0)

Finds the top motif for each subsequence length in the given range. The top motif is defined as the
distancematrix.valmod.find_variable_length_motifs(series, min_motif_length, max_motif_length, cache_size=3, noise_std=0.0)

Finds the top motif for each subsequence length in the given range. The top motif is defined as the subsequence (for a given length) for which the z-normalized euclidean distance is minimal, excluding any trivial matches.

This method implements the VALMOD algorithm described in “Matrix Profile X: VALMOD - Scalable Discovery of Variable-Length Motifs in Data Series” by M. Linardi et al.

Parameters

  • series – one dimensional time series

  • min_motif_length – minimum motif length

  • max_motif_length – maximum motif length (inclusive)

  • cache_size – number of entries kept in memory per subsequence (can only affect performance, default should be okay)

  • noise_std – standard deviation of noise on the signal, used for correcting the z-normalized euclidean distance

Returns

a list of tuples of length (max_motif_length - min_motif_length + 1), containing the indices of the motif and its match

class distancematrix.valmod.LowerBoundEntry(q_index, s_index, lower_bound_base, dot_prod)