tslearn-team / tslearn-team/tslearn
Is tslearn fully parallelizable for large-scale time series clustering?
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 3.2k
- Forks
- 384
- Avg merge
- 3d 12h
- Merged PRs (30d)
- 11
Description
Pleasure to see a parallellizable DTW metrics cdist_dtw is in tslearn. However, when the time series data set is very large, e.g., 20 GB, and run tslearn over a 50 cores server, both dtw and softdtw cannot fully utilize all 50 cores. It seems this feature is related to the implementation of TimeSeriesKMeans and the cdist_dtw, and cdist_softdtw. I am not quite sure about it and a discussion may be helpful.
Best regards,
Percy
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reading the implementations and documentation for TimeSeriesKMeans, cdist_dtw, and cdist_softdtw. Reproduce the reported workload on a multi-core system and inspect how parallelism is configured and where utilization drops. Done would require an agreed scope and evidence that the relevant clustering and distance computations use the available cores effectively.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning, performance
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 20/100