This is a supporting page to our paper -
Online Discovery and Maintenance of Time Series Motif


Abdullah Mueen and Eamonn Keogh
The paper is

Online Motif Discovery: Visual Demonstration

Code and Executables 

We have three versions of our algorithm. The versions are in compressed folders protected  by passwords. Please email an author for password.


1. Version 1: This version processes every real number in a dataset one by one from the file.  In a sense it assumes a very slow data rate. It is used to measure the performances (time and space usages) of our algorithm.

2.   Version 2: This version takes in a data rate and process the data in exactly this rate. If our algorithm falls behind the data then it skips subsequences.

3.   Version 3: If you are a practitioner and want to check out the motifs in your dataset, you should download this version which has beautiful MATLAB wrapper around the code and can play the data you have to show you the motifs in your data. For details on how the code works please see inside the folder.

    This Version also contains an adaptation for Mac OS X and Ubuntu which is done by Thomas Guyet and R. Quiniou.

4. Version 4: This is the ultimate tool for online motif discovery. We are working on it. Please check back later.


Spreadsheet of Experimental Resluts 

We have compiled results of all the experiments in a spreadsheet

Few things to note ...

  • All the real datasets are single time series
  • All the times are in seconds unless specified otherwise
  • All the distances are in z-normalized space.
  • All the codes use "Early Abandoning".


  • Synthetic data we use is a random walk (RW) dataset. The generator for this data is random_walk
  • EEG A down sampled version of this data can be found here.
  • EOG
  • InsectA and InsectB. We use a concatenation of the two if larger series is required.

Case Study: Online Summarization/Compression

For this case study we use ECG data.

Case Study: Acoustic Wildlife Management

The corresponding time series is here.

Case Study: Closing the Loop
For this case study we use the following data.

This page is created by -
Abdullah Mueen
Department of Computer Science and Engineering,
University of California - Riverside.