Backport PR #54884 on branch 2.1.x (REGR: value_counts raises with bi…

…ns) (#54932)

Backport PR #54884: REGR: value_counts raises with bins

Co-authored-by: Patrick Hoefler <61934744+phofl@users.noreply.github.com>

doc/source/whatsnew/v2.1.1.rst

@@ -15,6 +15,7 @@ Fixed regressions
 ~~~~~~~~~~~~~~~~~
 - Fixed regression in :func:`read_csv` when ``usecols`` is given and ``dtypes`` is a dict for ``engine="python"`` (:issue:`54868`)
 - Fixed regression in :meth:`DataFrame.__setitem__` raising ``AssertionError`` when setting a :class:`Series` with a partial :class:`MultiIndex` (:issue:`54875`)
+- Fixed regression in :meth:`Series.value_counts` raising for numeric data if ``bins`` was specified (:issue:`54857`)
 - Fixed regression when comparing a :class:`Series` with ``datetime64`` dtype with ``None`` (:issue:`54870`)
 
 .. ---------------------------------------------------------------------------

pandas/core/algorithms.py

@@ -878,7 +878,9 @@ def value_counts_internal(
     if bins is not None:
         from pandas.core.reshape.tile import cut
 
-        values = Series(values, copy=False)
+        if isinstance(values, Series):
+            values = values._values
+
         try:
             ii = cut(values, bins, include_lowest=True)
         except TypeError as err:

pandas/tests/test_algos.py

@@ -1412,6 +1412,19 @@ def test_value_counts_uint64(self):
 
         tm.assert_series_equal(result, expected)
 
+    def test_value_counts_series(self):
+        # GH#54857
+        values = np.array([3, 1, 2, 3, 4, np.nan])
+        result = Series(values).value_counts(bins=3)
+        expected = Series(
+            [2, 2, 1],
+            index=IntervalIndex.from_tuples(
+                [(0.996, 2.0), (2.0, 3.0), (3.0, 4.0)], dtype="interval[float64, right]"
+            ),
+            name="count",
+        )
+        tm.assert_series_equal(result, expected)
+
 
 class TestDuplicated:
     def test_duplicated_with_nas(self):