OneDAL python ifaces for kNN (#743)

* c++ code built

* c++ and python built and method called

* save work; python works, problem with indices

* fix problem with indices

* save work, feel free to comment

* fix patching

* light scope of tests were added

* add gpu support; dispatching for sklearnex

* apply part of comments

* save work, comments applied, patching for classification and unsupervised has been added

* fix for patching condition

* fix multioutput, y for unsupervised fit is none

* apply comment int32->64

* fix unsupervised patching, save work

* apply comment

* revert regression changes, codefactor fixes

* save work

* clear comments in test, need to add fresh tests

* fix unsupervised tests, warnings import

* apply @vlad-nazarov's comment

* remove redundant astype

* tests

* fixes for pep8

* fix codefactor

* fix pep8 hits concerning conditions

* applied comments

* fix condition

* update copyrights

* applied comment

* fix private ci

Author: KalyanovD

onedal/__init__.py

@@ -32,4 +32,4 @@
     import onedal._onedal_py_host as _backend
     _is_dpc_backend = False
 
-__all__ = ['primitives', 'svm']
+__all__ = ['neighbors', 'primitives', 'svm']

onedal/common/_estimator_checks.py

@@ -0,0 +1,43 @@
+#===============================================================================
+# Copyright 2022 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#===============================================================================
+
+
+def _check_is_fitted(estimator, attributes=None, *, msg=None):
+    if msg is None:
+        msg = ("This %(name)s instance is not fitted yet. Call 'fit' with "
+               "appropriate arguments before using this estimator.")
+
+    if not hasattr(estimator, 'fit'):
+        raise TypeError("%s is not an estimator instance." % (estimator))
+
+    if attributes is not None:
+        if not isinstance(attributes, (list, tuple)):
+            attributes = [attributes]
+        attrs = all([hasattr(estimator, attr) for attr in attributes])
+    else:
+        attrs = [v for v in vars(estimator)
+                 if v.endswith("_") and not v.startswith("__")]
+
+    if not attrs:
+        raise AttributeError(msg % {'name': type(estimator).__name__})
+
+
+def _is_classifier(estimator):
+    return getattr(estimator, "_estimator_type", None) == "classifier"
+
+
+def _is_regressor(estimator):
+    return getattr(estimator, "_estimator_type", None) == "regressor"

onedal/dal.cpp

@@ -34,6 +34,7 @@ ONEDAL_PY_INIT_MODULE(sigmoid_kernel);
 
 /* algorithms */
 ONEDAL_PY_INIT_MODULE(svm);
+ONEDAL_PY_INIT_MODULE(neighbors);
 
 #ifdef ONEDAL_DATA_PARALLEL
 PYBIND11_MODULE(_onedal_py_dpc, m) {
@@ -49,6 +50,7 @@ PYBIND11_MODULE(_onedal_py_host, m) {
     init_sigmoid_kernel(m);
 
     init_svm(m);
+    init_neighbors(m);
 }
 
 } // namespace oneapi::dal::python

onedal/datatypes/__init__.py

@@ -19,16 +19,17 @@
     _validate_targets,
     _check_X_y,
     _check_array,
-    _check_is_fitted,
     _check_classification_targets,
     _type_of_target,
     _is_integral_float,
     _is_multilabel,
     _check_n_features,
-    _num_features
+    _num_features,
+    _num_samples
 )
 
 __all__ = ['_column_or_1d', '_validate_targets', '_check_X_y',
-           '_check_array', '_check_is_fitted',
-           '_check_classification_targets', '_type_of_target', '_is_integral_float',
-           '_is_multilabel', '_check_n_features', '_num_features']
+           '_check_array', '_check_classification_targets',
+           '_type_of_target', '_is_integral_float',
+           '_is_multilabel', '_check_n_features', '_num_features',
+           '_num_samples']

onedal/datatypes/numpy_helpers.hpp

@@ -33,6 +33,10 @@
             _FUNCT(NPY_INT32);                              \
             break;                                          \
         }                                                   \
+        case dal::data_type::int64: {                       \
+            _FUNCT(NPY_INT64);                              \
+            break;                                          \
+        }                                                   \
         default: _EXCEPTION;                                \
     };
 
@@ -50,6 +54,10 @@
             _FUNCT(NPY_INT32, std::int32_t);                 \
             break;                                           \
         }                                                    \
+        case dal::data_type::int64: {                        \
+            _FUNCT(NPY_INT64, std::int64_t);                 \
+            break;                                           \
+        }                                                    \
         default: _EXCEPTION;                                 \
     };
 

onedal/datatypes/validation.py

@@ -20,6 +20,7 @@
 from scipy.sparse import issparse, dok_matrix, lil_matrix
 from scipy.sparse.base import spmatrix
 from collections.abc import Sequence
+from numbers import Integral
 
 
 class DataConversionWarning(UserWarning):
@@ -141,26 +142,6 @@ def _check_X_y(X, y, dtype="numeric", accept_sparse=False, order=None, copy=Fals
     return X, y
 
 
-def _check_is_fitted(estimator, attributes=None, *, msg=None):
-    if msg is None:
-        msg = ("This %(name)s instance is not fitted yet. Call 'fit' with "
-               "appropriate arguments before using this estimator.")
-
-    if not hasattr(estimator, 'fit'):
-        raise TypeError("%s is not an estimator instance." % (estimator))
-
-    if attributes is not None:
-        if not isinstance(attributes, (list, tuple)):
-            attributes = [attributes]
-        attrs = all([hasattr(estimator, attr) for attr in attributes])
-    else:
-        attrs = [v for v in vars(estimator)
-                 if v.endswith("_") and not v.startswith("__")]
-
-    if not attrs:
-        raise AttributeError(msg % {'name': type(estimator).__name__})
-
-
 def _check_classification_targets(y):
     y_type = _type_of_target(y)
     if y_type not in ['binary', 'multiclass', 'multiclass-multioutput',
@@ -329,3 +310,31 @@ def _num_features(X):
         return len(first_sample)
     except Exception as err:
         raise TypeError(message) from err
+
+
+def _num_samples(x):
+    message = "Expected sequence or array-like, got %s" % type(x)
+    if hasattr(x, "fit") and callable(x.fit):
+        # Don't get num_samples from an ensembles length!
+        raise TypeError(message)
+
+    if not hasattr(x, "__len__") and not hasattr(x, "shape"):
+        if hasattr(x, "__array__"):
+            x = np.asarray(x)
+        else:
+            raise TypeError(message)
+
+    if hasattr(x, "shape") and x.shape is not None:
+        if len(x.shape) == 0:
+            raise TypeError(
+                "Singleton array %r cannot be considered a valid collection." % x
+            )
+    # Check that shape is returning an integer or default to len
+    # Dask dataframes may not return numeric shape[0] value
+    if isinstance(x.shape[0], Integral):
+        return x.shape[0]
+
+    try:
+        return len(x)
+    except TypeError as type_error:
+        raise TypeError(message) from type_error

onedal/neighbors/__init__.py

@@ -0,0 +1,19 @@
+#===============================================================================
+# Copyright 2022 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#===============================================================================
+
+from .neighbors import KNeighborsClassifier, NearestNeighbors
+
+__all__ = ['KNeighborsClassifier', 'NearestNeighbors']