-
The CountVectorizer default dtype is `numpy.int64`.
Since there are no negative values in count vectors, perhaps it would be better to use `numpy.uint64` as the default?
-
I understand why hash seems to be a better solution for distributed text preprocessing, but I also need a way to make my features human-readable. It seems like spark has a [CountVectorizer](https://sp…
-
#### Description
clang: error: unsupported option '-fopenmp'
#### Steps/Code to Reproduce
executed command: pip install scikit-learn
#### Expected Results
#### Actual Results
…
-
#### Describe the bug
When I used LassoCV with `fit_intercept=False` and `precompute=np.dot(X.T, X)`, the result was different from the case without `precompute`.
#### Steps/Code to Reprod…
-
Great work, Kiran! You now have a solid project that you've completed from start to finish.
Here is my feedback so far:
**General suggestions**
1. I suggest you look at the following style gui…
-
#### Describe the bug
'SVC' object has no attribute '_probA'. While using the SVM classifier from the sklearn package I encountered this result. I had trained my model on a dataset and then sav…
-
```python
In [1]: import dask.bag as db
In [2]: import dask_ml.feature_extraction.text
In [3]: from dask.distributed import Client
...: client = Client()
In [4]: vocab = {"foo": 0, "bar": …
-
[Rasa](https://rasa.com/docs/rasa/nlu/components/#countvectorsfeaturizer) adds an extra property to the countvectorizer for out of vocabulary words. This can be very useful in outlier detection of tex…
-
#### Description
The documentation of default values in many classes is either not included, inconsistent in how it is written, or out-of-date. I would like to gather a few people to work on the defa…
-
I'd want to be able to do TF-IDF calculations in Dask in a similar fashion as in apache spark. https://spark.apache.org/docs/latest/mllib-feature-extraction.html
Right now, counting/hashing vectori…