[REVIEW] Fix cuDF to cuPy conversion (missing value)

CHANGELOG.md

@@ -56,6 +56,7 @@
 - PR #3008: Check number of columns in check_array validator
 - PR #3012: Increasing learning rate for SGD log loss and invscaling pytests
 - PR #2950: Fix includes in UMAP
+- PR #3194: Fix cuDF to cuPy conversion (missing value)
 - PR #3021: Fix a hang in cuML RF experimental backend
 - PR #3039: Update RF and decision tree parameter initializations in benchmark codes
 - PR #3060: Speed up test suite `test_fil`

python/cuml/common/input_utils.py

@@ -398,11 +398,21 @@ def input_to_cupy_array(X,
                         check_cols=False,
                         check_rows=False,
                         fail_on_order=False,
-                        force_contiguous=True) -> cuml_array:
+                        force_contiguous=True,
+                        fail_on_nan=True) -> cuml_array:
     """
     Identical to input_to_cuml_array but it returns a cupy array instead of
     CumlArray
     """
+    if not fail_on_nan:
+        if isinstance(X, (cudf.DataFrame, cudf.Series)):
+            try:
+                X = X.values
+            except ValueError:
+                X = X.astype('float64', copy=False)
+                X.fillna(cp.nan, inplace=True)
+                X = X.values
+
     out_data = input_to_cuml_array(X,
                                    order=order,
                                    deepcopy=deepcopy,

python/cuml/test/test_input_utils.py

@@ -23,6 +23,7 @@
 
 from cuml.common import input_to_cuml_array, CumlArray
 from cuml.common import input_to_host_array
+from cuml.common.input_utils import input_to_cupy_array
 from cuml.common import has_cupy
 from cuml.common.input_utils import convert_dtype
 from cuml.common.memory_utils import _check_array_contiguity
@@ -344,3 +345,26 @@ def get_input(type, nrows, ncols, dtype, order='C', out_dtype=False):
                                 order=order)
     else:
         return result, np.array(cp.asnumpy(rand_mat), order=order)
+
+
+def test_tocupy_missing_values_handling():
+    df = cudf.DataFrame(data=[[7, 2, 3], [4, 5, 6], [10, 5, 9]])
+    array, n_rows, n_cols, dtype = input_to_cupy_array(df, fail_on_nan=False)
+    assert isinstance(array, cp.ndarray)
+    assert str(array.dtype) == 'int64'
+
+    df = cudf.DataFrame(data=[[7, 2, 3], [4, None, 6], [10, 5, 9]])
+    array, n_rows, n_cols, dtype = input_to_cupy_array(df, fail_on_nan=False)
+    assert isinstance(array, cp.ndarray)
+    assert str(array.dtype) == 'float64'
+    assert cp.isnan(array[1, 1])
+
+    df = cudf.Series(data=[7, None, 3])
+    array, n_rows, n_cols, dtype = input_to_cupy_array(df, fail_on_nan=False)
+    assert str(array.dtype) == 'float64'
+    assert cp.isnan(array[1])
+
+    with pytest.raises(ValueError):
+        df = cudf.Series(data=[7, None, 3])
+        array, n_rows, n_cols, dtype = input_to_cupy_array(df,
+                                                           fail_on_nan=True)

python/cuml/thirdparty_adapters/adapters.py

@@ -277,7 +277,8 @@ def check_array(array, accept_sparse=False, accept_large_sparse=True,
     else:
         X, n_rows, n_cols, dtype = input_to_cupy_array(array,
                                                        order=order,
-                                                       deepcopy=copy)
+                                                       deepcopy=copy,
+                                                       fail_on_nan=False)
         if correct_dtype != dtype:
             X = X.astype(correct_dtype)
         check_finite(X, force_all_finite)