Unable to reproduce OneHotEncoder example from the docs #685
Description
This example from the API reference returns an error:
from dask_ml.preprocessing import OneHotEncoder
import numpy as np
import dask.array as da
enc = OneHotEncoder()
X = da.from_array(np.array([['A'], ['B'], ['A'], ['C']]), chunks=2)
enc.fit(X)
enc.categories_
enc.transform(X)This is the traceback:
Details
ValueErrorTraceback (most recent call last)
<ipython-input-1-f54891b18539> in <module>
6 enc.fit(X)
7 enc.categories_
----> 8 enc.transform(X)
~/myenv/lib/python3.7/site-packages/dask_ml/preprocessing/_encoders.py in transform(self, X)
211 self, X: Union[ArrayLike, DataFrameType]
212 ) -> Union[ArrayLike, DataFrameType]:
--> 213 return self._transform(X)
214
215 def _transform_new(
~/myenv/lib/python3.7/site-packages/dask_ml/preprocessing/_encoders.py in _transform(self, X, handle_unknown)
243 for i in range(n_features)
244 ]
--> 245 X = da.concatenate(Xs, axis=1)
246
247 if not self.sparse:
~/myenv/lib/python3.7/site-packages/dask/array/core.py in concatenate(seq, axis, allow_unknown_chunksizes)
3480 raise ValueError("Need array(s) to concatenate")
3481
-> 3482 meta = np.concatenate([meta_from_array(s) for s in seq], axis=axis)
3483
3484 # Promote types to match meta
<__array_function__ internals> in concatenate(*args, **kwargs)
ValueError: zero-dimensional arrays cannot be concatenatedEnvironment:
- Dask version: 2.19.0
- numpy version: 1.18.5
- Python version: 3.7.6
- Install method (conda, pip, source): conda