Impl Series.skew() (#813)

Author: Rubtsowa

examples/series/series_skew.py

@@ -0,0 +1,39 @@
+# *****************************************************************************
+# Copyright (c) 2020, Intel Corporation All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+#     Redistributions of source code must retain the above copyright notice,
+#     this list of conditions and the following disclaimer.
+#
+#     Redistributions in binary form must reproduce the above copyright notice,
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+# THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+# OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+# WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+# OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
+# EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# *****************************************************************************
+
+import pandas as pd
+import numpy as np
+from numba import njit
+
+
+@njit
+def series_skew():
+    s = pd.Series([np.nan, -2., 3., 5.0])
+
+    return s.skew()  # Expect -1.1520696383139375
+
+
+print(series_skew())

sdc/datatypes/hpat_pandas_series_functions.py

@@ -4818,3 +4818,67 @@ def sdc_pandas_series_groupby_impl(self, by=None, axis=0, level=None, as_index=T
         return init_series_groupby(self, by, grouped, sort)
 
     return sdc_pandas_series_groupby_impl
+
+
+@sdc_overload_method(SeriesType, 'skew')
+def sdc_pandas_series_skew(self, axis=None, skipna=None, level=None, numeric_only=None):
+    """
+        Intel Scalable Dataframe Compiler User Guide
+        ********************************************
+
+        Pandas API: pandas.Series.skew
+
+        Limitations
+        -----------
+        - Parameters ``level`` and ``numeric_only`` are supported only with default value ``None``.
+
+        Examples
+        --------
+        .. literalinclude:: ../../../examples/series/series_skew.py
+           :language: python
+           :lines: 27-
+           :caption: Unbiased rolling skewness.
+           :name: ex_series_skew
+
+        .. command-output:: python ./series/series_skew.py
+           :cwd: ../../../examples
+
+        Intel Scalable Dataframe Compiler Developer Guide
+        *************************************************
+        Pandas Series method :meth:`pandas.Series.skew` implementation.
+
+        .. only:: developer
+            Test: python -m sdc.runtests -k sdc.tests.test_series.TestSeries.test_series_skew*
+        """
+    _func_name = 'Method Series.skew()'
+
+    ty_checker = TypeChecker(_func_name)
+    ty_checker.check(self, SeriesType)
+
+    if not isinstance(axis, (types.Integer, types.NoneType, types.Omitted)) and axis is not None:
+        ty_checker.raise_exc(axis, 'int64', 'axis')
+
+    if not isinstance(skipna, (types.Boolean, types.NoneType, types.Omitted)) and skipna is not None:
+        ty_checker.raise_exc(skipna, 'bool', 'skipna')
+
+    if not isinstance(level, (types.Omitted, types.NoneType)) and level is not None:
+        ty_checker.raise_exc(level, 'None', 'level')
+
+    if not isinstance(numeric_only, (types.Omitted, types.NoneType)) and numeric_only is not None:
+        ty_checker.raise_exc(numeric_only, 'None', 'numeric_only')
+
+    def sdc_pandas_series_skew_impl(self, axis=None, skipna=None, level=None, numeric_only=None):
+        if axis != 0 and axis is not None:
+            raise ValueError('Parameter axis must be only 0 or None.')
+
+        if skipna is None:
+            _skipna = True
+        else:
+            _skipna = skipna
+
+        if _skipna:
+            return numpy_like.nanskew(self._data)
+
+        return numpy_like.skew(self._data)
+
+    return sdc_pandas_series_skew_impl

sdc/datatypes/hpat_pandas_series_rolling_functions.py

@@ -36,6 +36,7 @@
 
 from sdc.datatypes.common_functions import _almost_equal
 from sdc.datatypes.hpat_pandas_series_rolling_types import SeriesRollingType
+from sdc.functions.statistics import skew_formula
 from sdc.hiframes.pd_series_type import SeriesType
 from sdc.utilities.prange_utils import parallel_chunks
 from sdc.utilities.sdc_typing_utils import TypeChecker
@@ -549,15 +550,7 @@ def skew_result_or_nan(nfinite, minp, result):
 
     _sum, square_sum, cube_sum = result
 
-    n = nfinite
-    m2 = (square_sum - _sum * _sum / n) / n
-    m3 = (cube_sum - 3.*_sum*square_sum/n + 2.*_sum*_sum*_sum/n/n) / n
-    res = 0 if m2 == 0 else m3 / m2 ** 1.5
-
-    if (n > 2) & (m2 > 0):
-        res = numpy.sqrt((n - 1.) * n) / (n - 2.) * m3 / m2 ** 1.5
-
-    return res
+    return skew_formula(nfinite, _sum, square_sum, cube_sum)
 
 
 @sdc_register_jitable

sdc/functions/numpy_like.py

@@ -44,6 +44,7 @@
 from numba.typed import List
 
 import sdc
+from sdc.functions.statistics import skew_formula
 from sdc.utilities.sdc_typing_utils import TypeChecker
 from sdc.utilities.utils import (sdc_overload, sdc_register_jitable,
                                  min_dtype_int_val, max_dtype_int_val, min_dtype_float_val,
@@ -988,3 +989,65 @@ def getitem_by_mask_impl(arr, idx):
             return result_data
 
     return getitem_by_mask_impl
+
+
+def skew(a):
+    pass
+
+
+def nanskew(a):
+    pass
+
+
+@sdc_overload(skew)
+def np_skew(arr):
+    if not isinstance(arr, types.Array):
+        return
+
+    def skew_impl(arr):
+        len_val = len(arr)
+        n = 0
+        _sum = 0.
+        square_sum = 0.
+        cube_sum = 0.
+
+        for idx in numba.prange(len_val):
+            if not numpy.isnan(arr[idx]):
+                n += 1
+                _sum += arr[idx]
+                square_sum += arr[idx] ** 2
+                cube_sum += arr[idx] ** 3
+
+        if n == 0 or n < len_val:
+            return numpy.nan
+
+        return skew_formula(n, _sum, square_sum, cube_sum)
+
+    return skew_impl
+
+
+@sdc_overload(nanskew)
+def np_nanskew(arr):
+    if not isinstance(arr, types.Array):
+        return
+
+    def nanskew_impl(arr):
+        len_val = len(arr)
+        n = 0
+        _sum = 0.
+        square_sum = 0.
+        cube_sum = 0.
+
+        for idx in numba.prange(len_val):
+            if not numpy.isnan(arr[idx]):
+                n += 1
+                _sum += arr[idx]
+                square_sum += arr[idx] ** 2
+                cube_sum += arr[idx] ** 3
+
+        if n == 0:
+            return numpy.nan
+
+        return skew_formula(n, _sum, square_sum, cube_sum)
+
+    return nanskew_impl

sdc/functions/statistics.py

@@ -0,0 +1,41 @@
+# -*- coding: utf-8 -*-
+# *****************************************************************************
+# Copyright (c) 2020, Intel Corporation All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+#     Redistributions of source code must retain the above copyright notice,
+#     this list of conditions and the following disclaimer.
+#
+#     Redistributions in binary form must reproduce the above copyright notice,
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
+# THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+# OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+# WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+# OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
+# EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+# *****************************************************************************
+
+import numpy
+from sdc.utilities.utils import sdc_register_jitable
+
+
+@sdc_register_jitable
+def skew_formula(n, _sum, square_sum, cube_sum):
+    m2 = (square_sum - _sum * _sum / n) / n
+    m3 = (cube_sum - 3. * _sum * square_sum / n + 2. * _sum * _sum * _sum / n / n) / n
+    res = numpy.nan if m2 == 0 else m3 / m2 ** 1.5
+
+    if (n > 2) & (m2 > 0):
+        res = numpy.sqrt((n - 1.) * n) / (n - 2.) * m3 / m2 ** 1.5
+
+    return res

sdc/tests/test_series.py

@@ -7188,6 +7188,73 @@ def test_impl(A, idx):
                 msg = 'The index of boolean indexer is not comparable to Series index.'
                 self.assertIn(msg, str(raises.exception))
 
+    def test_series_skew(self):
+        def test_impl(series, axis, skipna):
+            return series.skew(axis=axis, skipna=skipna)
+
+        hpat_func = self.jit(test_impl)
+        test_data = [[6, 6, 2, 1, 3, 3, 2, 1, 2],
+                     [1.1, 0.3, 2.1, 1, 3, 0.3, 2.1, 1.1, 2.2],
+                     [6, 6.1, 2.2, 1, 3, 3, 2.2, 1, 2],
+                     [],
+                     [6, 6, np.nan, 2, np.nan, 1, 3, 3, np.inf, 2, 1, 2, np.inf],
+                     [1.1, 0.3, np.nan, 1.0, np.inf, 0.3, 2.1, np.nan, 2.2, np.inf],
+                     [1.1, 0.3, np.nan, 1, np.inf, 0, 1.1, np.nan, 2.2, np.inf, 2, 2],
+                     [np.nan, np.nan, np.nan],
+                     [np.nan, np.nan, np.inf],
+                     [np.inf, 0, np.inf, 1, 2, 3, 4, 5]
+                     ]
+        all_test_data = test_data + test_global_input_data_float64
+        for data in all_test_data:
+            with self.subTest(data=data):
+                s = pd.Series(data)
+                for axis in [0, None]:
+                    with self.subTest(axis=axis):
+                        for skipna in [None, False, True]:
+                            with self.subTest(skipna=skipna):
+                                res1 = test_impl(s, axis, skipna)
+                                res2 = hpat_func(s, axis, skipna)
+                                np.testing.assert_allclose(res1, res2)
+
+    def test_series_skew_default(self):
+        def test_impl():
+            s = pd.Series([np.nan, -2., 3., 9.1])
+            return s.skew()
+
+        hpat_func = self.jit(test_impl)
+        np.testing.assert_allclose(test_impl(), hpat_func())
+
+    def test_series_skew_not_supported(self):
+        def test_impl(series, axis=None, skipna=None, level=None, numeric_only=None):
+            return series.skew(axis=axis, skipna=skipna, level=level, numeric_only=numeric_only)
+
+        hpat_func = self.jit(test_impl)
+        s = pd.Series([1.1, 0.3, np.nan, 1, np.inf, 0, 1.1, np.nan, 2.2, np.inf, 2, 2])
+        with self.assertRaises(TypingError) as raises:
+            hpat_func(s, axis=0.75)
+        msg = 'TypingError: Method Series.skew() The object axis\n given: float64\n expected: int64'
+        self.assertIn(msg, str(raises.exception))
+
+        with self.assertRaises(TypingError) as raises:
+            hpat_func(s, skipna=0)
+        msg = 'TypingError: Method Series.skew() The object skipna\n given: int64\n expected: bool'
+        self.assertIn(msg, str(raises.exception))
+
+        with self.assertRaises(TypingError) as raises:
+            hpat_func(s, level=0)
+        msg = 'TypingError: Method Series.skew() The object level\n given: int64\n expected: None'
+        self.assertIn(msg, str(raises.exception))
+
+        with self.assertRaises(TypingError) as raises:
+            hpat_func(s, numeric_only=0)
+        msg = 'TypingError: Method Series.skew() The object numeric_only\n given: int64\n expected: None'
+        self.assertIn(msg, str(raises.exception))
+
+        with self.assertRaises(ValueError) as raises:
+            hpat_func(s, axis=5)
+        msg = 'Parameter axis must be only 0 or None.'
+        self.assertIn(msg, str(raises.exception))
+
 
 if __name__ == "__main__":
     unittest.main()

sdc/tests/tests_perf/test_perf_series.py

@@ -137,6 +137,7 @@ def _test_case(self, pyfunc, name, total_data_length, input_data=None, data_num=
     TC(name='shape', size=[10 ** 7], call_expr='data.shape', usecase_params='data'),
     TC(name='shift', size=[10 ** 8]),
     TC(name='size', size=[10 ** 7], call_expr='data.size', usecase_params='data'),
+    TC(name='skew', size=[10 ** 8], check_skipna=True),
     TC(name='sort_values', size=[10 ** 5]),
     TC(name='std', size=[10 ** 7], check_skipna=True),
     TC(name='sub', size=[10 ** 7], params='other', data_num=2),