Are umap transformations non-deterministic?
Open
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 8.3k
- Forks
- 871
- Avg merge
- 1d 13h
- Merged PRs (30d)
- 5
Description
I am trying to use umap to preprocess some data, and I've noticed that the same vector gives a different result according to the number of rows that is being passed to the transformation.
i.e the same row vector A outputs different vector according to the shape of the data (# of rows) being transformed.
# fit umap to data X
reducer = umap.UMAP().fit(X)
# transform X using reducer
embedding = reducer.transform(X)
# get subset of X to transform
embedding_sub = reducer.transform(X[:100,:])
# => I was assuming embedding_sub == embedding[:100, :]
# => but that wasn't the case
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 reproducing the issue with the provided fit and transform calls, comparing transform(X) with transform(X[:100, :]). Then inspect the transformation entry point and related tests to determine whether identical rows should produce identical outputs regardless of batch size; done means the behavior is explained or covered by a confirmed fix.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100