Fix tests (#965)

* text matrix

* spliting the string creates the expected input to FeatureHasher #964

* FeatureHasher issue #963

* addressing catagories_ type mismatch when auto by explicitly setting dtype on test data to object #964

* reverted to just ubuntu for time saving

.github/workflows/tests.yaml

@@ -41,4 +41,4 @@ jobs:
 
       - name: Run tests
         shell: bash -l {0}
-        run: pytest -v
+        run: pytest -v

.gitignore

@@ -94,6 +94,9 @@ ENV/
 .spyderproject
 .spyproject
 
+# VS Code
+*.code-workspace
+
 # Rope project settings
 .ropeproject
 

tests/feature_extraction/test_text.py

@@ -32,7 +32,9 @@
 )
 def test_basic(vect, container):
     b = db.from_sequence(JUNK_FOOD_DOCS, npartitions=2)
-    if container == "series":
+    if type(vect) == dask_ml.feature_extraction.text.FeatureHasher:
+        b = b.str.split()
+    elif container == "series":
         b = b.to_dataframe(columns=["text"])["text"]
     elif container == "array":
         b = b.to_dataframe(columns=["text"])["text"].values

tests/preprocessing/test_encoders.py

@@ -11,7 +11,7 @@
 from dask_ml._compat import DASK_2_20_0, PANDAS_VERSION
 from dask_ml.utils import assert_estimator_equal
 
-X = np.array([["a"], ["a"], ["b"], ["c"]])
+X = np.array([["a"], ["a"], ["b"], ["c"]], dtype=object)
 dX = da.from_array(X, 2)
 df = pd.DataFrame(X, columns=["A"]).apply(lambda x: x.astype("category"))
 ddf = dd.from_pandas(df, npartitions=2)