Python implementation of k-means

k-means in Python using dask arrays, and init function from dask-ml.

added 4 commits

• ee4a2ba1 - KMeansMachine as Transformer
• e40c0ea7 - Tests for python k-means
• 510e8517 - Plot script to use python k-means
• 4c21f381 - Update the cluster module init

Compare with previous version

assigned to @amohammadi and unassigned @ydayer

added 1 commit

• 475fda6e - Add dask and dask-ml to requirements

Compare with previous version

added 2 commits

• 75543d3b - remove dask from requirements.txt to fix a possible issue
• 0356d983 - Improve tests error messages

Compare with previous version

added 2 commits

• ada566f6 - remove test_kmeans_trainer.py
• 3efa9662 - clean-up the tests

Compare with previous version

I think we may need to revisit this:

    def fit(self, X, y=None, trainer=None):

now or in the future.

I think when you put this estimator in a scikit-learn pipeline, then you can no longer change the trainer. So this might be an issue.

merged

mentioned in commit 8119eed1

mentioned in commit 6f4a29cb

mentioned in issue bob#269 (closed)