pyattimo 0.5.0

A fast and scalable algorithm for time series motif mining.

This is a python wrapper for ATTIMO, a fast algorithm for mining time series motifs, with probabilistic guarantees.

The inner workings and guarantees of the algorithm are described in this paper.

If you find this software useful for your research, please use the following citation:

@article{DBLP:journals/pvldb/CeccarelloG22,
  author    = {Matteo Ceccarello and
               Johann Gamper},
  title     = {Fast and Scalable Mining of Time Series Motifs with Probabilistic
               Guarantees},
  journal   = {Proc. {VLDB} Endow.},
  volume    = {15},
  number    = {13},
  pages     = {3841--3853},
  year      = {2022},
  url       = {https://www.vldb.org/pvldb/vol15/p3841-ceccarello.pdf},
  timestamp = {Wed, 11 Jan 2023 17:06:38 +0100},
  biburl    = {https://dblp.org/rec/journals/pvldb/CeccarelloG22.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}

Installation

pyATTIMO is a Rust library wrapped in Python. Therefore, if a wheel is available for your platform, you can install it simply by invoking:

pip install pyattimo

Otherwise, you need the Rust toolchain installed to be able to compile it. The simplest way is to visit https://rustup.rs/ and follow the instructions there. You will need the nightly toolchain:

curl https://sh.rustup.rs -sSf | sh -s -- --default-toolchain nightly

After that, you can just run:

pip install pyattimo

At this point, you should have the pyattimo library available in your interpreter.

Usage

In essence, the library provides an iterator over the motifs of the given time series. The following snippet illustrates the basic usage:

import pyattimo

# Load an example time series
ts = pyattimo.load_dataset("ecg", prefix=1000000)

# Create the motifs iterator
motifs = pyattimo.MotifsIterator(ts, w=1000, max_k=100)

# Get the top motif via the iterator interface
m = next(motifs)

# Plot the motif just obtained
m.plot()

Further information and examples can be found here