Demo

You can easily try this yourself.

1. Install COBRAS-TS with the visual user interface:
   pip install cobras_ts[gui]
2. Run COBRAS-TS on a sample of the CBF dataset from the UCR archive
   cobras_ts --visual --format=csv --dist=kshape --labelcol=0 https://raw.githubusercontent.com/ML-KULeuven/cobras/master/cobras_ts/cli/webapp/data/CBF_TEST_SAMPLE

Installation

COBRAS is available on PyPi, install it using:
pip install cobras_ts
This software can be run from the commandline or used as a Python package, examples can be found on GitHub.
If you want to run COBRAS-TS with the visual user interface, use
pip install cobras_ts[gui]

For more details on installation and usage see the GitHub repository.

Code

COBRA is available on BitBucket and COBRAS(-TS) is available on GitHub.

Papers

• Van Craenendonck, T., Meert, W., Dumancic, S. & Blockeel, H. Interactive Time Series Clustering with COBRAS-TS. Demo track of ECMLPKDD, 2018. (PDF,CODE)

  
  Demo paper introducing a web application
  for interactive clustering of time series based on COBRAS-TS.

---

• Van Craenendonck, T., Meert, W., Dumancic, S. & Blockeel, H. COBRAS-TS: A new approach to Semi-Supervised Clustering of Time Series . Discovery Science, 2018. (PDF, CODE, SLIDES)
  
  

  Best paper award
  COBRAS-TS extends COBRAS to time series clustering, demonstrating for the first time that time series clustering can benefit greatly from small amounts of user interaction.

---

• Van Craenendonck, T., Dumancic, S., Vanwolputte, E. & Blockeel, H. COBRAS: Interactive clustering with pairwise constraints. Intelligent Data Analysis, 2018. (PDF, CODE, SLIDES)
  
  

  COBRAS dynamically refines super-instances during clustering,
  making it suitable for interactive clustering.

---

• Van Craenendonck, T., Dumancic, S. & Blockeel, H. COBRA: A fast and simple method for active clustering with pairwise constraints. International Joint Conference on Artificial Intelligence, 2017. (Lirias,PDF, CODE, SLIDES)
  
  

  The paper that introduces the concept of super-instances to semi-supervised clustering.

Contact

For any questions about the paper, code, ... contact Toon Van Craenendonck at toonvancraenendonck@gmail.com.