bump min deps for 0.15

ci/requirements/doc.yml

@@ -17,7 +17,7 @@ dependencies:
   - numba
   - numpy
   - numpydoc
-  - pandas<0.25  # Hack around https://github.com/pydata/xarray/issues/3369
+  - pandas
   - rasterio
   - seaborn
   - sphinx

ci/requirements/py36-bare-minimum.yml

@@ -7,5 +7,5 @@ dependencies:
   - pytest
   - pytest-cov
   - pytest-env
-  - numpy=1.14
-  - pandas=0.24
+  - numpy=1.15
+  - pandas=0.25

ci/requirements/py36-min-all-deps.yml

@@ -29,8 +29,8 @@ dependencies:
   - nc-time-axis=1.2
   - netcdf4=1.4
   - numba=0.44
-  - numpy=1.14
-  - pandas=0.24
+  - numpy=1.15
+  - pandas=0.25
   # - pint  # See py36-min-nep18.yml
   - pip
   - pseudonetcdf=3.0
@@ -40,7 +40,7 @@ dependencies:
   - pytest-cov
   - pytest-env
   - rasterio=1.0
-  - scipy=1.0  # Policy allows for 1.2, but scipy>=1.1 breaks numpy=1.14
+  - scipy=1.2  # Policy allows for 1.3
   - seaborn=0.9
   # - sparse  # See py36-min-nep18.yml
   - toolz=0.10

ci/requirements/py36-min-nep18.yml

@@ -9,7 +9,7 @@ dependencies:
   - dask=2.4
   - distributed=2.4
   - numpy=1.17
-  - pandas=0.24
+  - pandas=0.25
   - pint=0.9  # Actually not enough as it doesn't implement __array_function__yet!
   - pytest
   - pytest-cov

doc/installing.rst

@@ -7,8 +7,8 @@ Required dependencies
 ---------------------
 
 - Python (3.6 or later)
-- `numpy <http://www.numpy.org/>`__ (1.14 or later)
-- `pandas <http://pandas.pydata.org/>`__ (0.24 or later)
+- `numpy <http://www.numpy.org/>`__ (1.15 or later)
+- `pandas <http://pandas.pydata.org/>`__ (0.25 or later)
 
 Optional dependencies
 ---------------------

doc/whats-new.rst

@@ -21,6 +21,15 @@ v0.15.0 (unreleased)
 
 Breaking changes
 ~~~~~~~~~~~~~~~~
+- Bumped minimum tested versions for dependencies:
+  - numpy 1.15
+  - pandas 0.25
+  - scipy 1.3
+  - bottleneck 1.3
+  - h5py 2.10
+  - rasterio 1.1
+  - pseudonetcdf
+
 - Remove ``compat`` and ``encoding`` kwargs from ``DataArray``, which
   have been deprecated since 0.12. (:pull:`3650`). 
   Instead, specify the encoding when writing to disk or set 

setup.py

@@ -22,7 +22,7 @@
 ]
 
 PYTHON_REQUIRES = ">=3.6"
-INSTALL_REQUIRES = ["numpy >= 1.14", "pandas >= 0.24"]
+INSTALL_REQUIRES = ["numpy >= 1.15", "pandas >= 0.25"]
 needs_pytest = {"pytest", "test", "ptr"}.intersection(sys.argv)
 SETUP_REQUIRES = ["pytest-runner >= 4.2"] if needs_pytest else []
 TESTS_REQUIRE = ["pytest >= 2.7.1"]

xarray/core/dataarray.py

@@ -2969,7 +2969,7 @@ def quantile(
 
         See Also
         --------
-        numpy.nanpercentile, pandas.Series.quantile, Dataset.quantile
+        numpy.nanquantile, pandas.Series.quantile, Dataset.quantile
 
         Examples
         --------

xarray/core/dataset.py

@@ -4466,11 +4466,7 @@ def _set_sparse_data_from_dataframe(
 
         idx = dataframe.index
         if isinstance(idx, pd.MultiIndex):
-            try:
-                codes = idx.codes
-            except AttributeError:
-                # deprecated since pandas 0.24
-                codes = idx.labels
+            codes = idx.codes
             coords = np.stack([np.asarray(code) for code in codes], axis=0)
             is_sorted = idx.is_lexsorted
         else:
@@ -5171,7 +5167,7 @@ def quantile(
 
         See Also
         --------
-        numpy.nanpercentile, pandas.Series.quantile, DataArray.quantile
+        numpy.nanquantile, pandas.Series.quantile, DataArray.quantile
 
         Examples
         --------

xarray/core/groupby.py

@@ -595,7 +595,7 @@ def quantile(self, q, dim=None, interpolation="linear", keep_attrs=None):
 
         See Also
         --------
-        numpy.nanpercentile, pandas.Series.quantile, Dataset.quantile,
+        numpy.nanquantile, pandas.Series.quantile, Dataset.quantile,
         DataArray.quantile
 
         Examples

xarray/core/nputils.py

@@ -14,12 +14,7 @@
 
 
 def _validate_axis(data, axis):
-    ndim = data.ndim
-    if not -ndim <= axis < ndim:
-        raise IndexError(f"axis {axis!r} out of bounds [-{ndim}, {ndim})")
-    if axis < 0:
-        axis += ndim
-    return axis
+    return np.core.multiarray.normalize_axis_index(axis, data.ndim)
 
 
 def _select_along_axis(values, idx, axis):

xarray/core/variable.py

@@ -1722,7 +1722,7 @@ def quantile(self, q, dim=None, interpolation="linear", keep_attrs=None):
 
         See Also
         --------
-        numpy.nanpercentile, pandas.Series.quantile, Dataset.quantile,
+        numpy.nanquantile, pandas.Series.quantile, Dataset.quantile,
         DataArray.quantile
         """
 
@@ -1734,10 +1734,6 @@ def quantile(self, q, dim=None, interpolation="linear", keep_attrs=None):
         scalar = utils.is_scalar(q)
         q = np.atleast_1d(np.asarray(q, dtype=np.float64))
 
-        # TODO: remove once numpy >= 1.15.0 is the minimum requirement
-        if np.count_nonzero(q < 0.0) or np.count_nonzero(q > 1.0):
-            raise ValueError("Quantiles must be in the range [0, 1]")
-
         if dim is None:
             dim = self.dims
 
@@ -1746,9 +1742,7 @@ def quantile(self, q, dim=None, interpolation="linear", keep_attrs=None):
 
         def _wrapper(npa, **kwargs):
             # move quantile axis to end. required for apply_ufunc
-
-            # TODO: use np.nanquantile once numpy >= 1.15.0 is the minimum requirement
-            return np.moveaxis(np.nanpercentile(npa, **kwargs), 0, -1)
+            return np.moveaxis(np.nanquantile(npa, **kwargs), 0, -1)
 
         axis = np.arange(-1, -1 * len(dim) - 1, -1)
         result = apply_ufunc(
@@ -1760,7 +1754,7 @@ def _wrapper(npa, **kwargs):
             output_dtypes=[np.float64],
             output_sizes={"quantile": len(q)},
             dask="parallelized",
-            kwargs={"q": q * 100, "axis": axis, "interpolation": interpolation},
+            kwargs={"q": q, "axis": axis, "interpolation": interpolation},
         )
 
         # for backward compatibility

xarray/plot/utils.py

@@ -547,10 +547,6 @@ def _add_colorbar(primitive, ax, cbar_ax, cbar_kwargs, cmap_params):
 
 def _rescale_imshow_rgb(darray, vmin, vmax, robust):
     assert robust or vmin is not None or vmax is not None
-    # TODO: remove when min numpy version is bumped to 1.13
-    # There's a cyclic dependency via DataArray, so we can't import from
-    # xarray.ufuncs in global scope.
-    from xarray.ufuncs import maximum, minimum
 
     # Calculate vmin and vmax automatically for `robust=True`
     if robust:
@@ -579,9 +575,7 @@ def _rescale_imshow_rgb(darray, vmin, vmax, robust):
     # After scaling, downcast to 32-bit float.  This substantially reduces
     # memory usage after we hand `darray` off to matplotlib.
     darray = ((darray.astype("f8") - vmin) / (vmax - vmin)).astype("f4")
-    with warnings.catch_warnings():
-        warnings.filterwarnings("ignore", "xarray.ufuncs", PendingDeprecationWarning)
-        return minimum(maximum(darray, 0), 1)
+    return np.minimum(np.maximum(darray, 0), 1)
 
 
 def _update_axes(

xarray/tests/test_dataarray.py

@@ -79,10 +79,6 @@ def test_repr_multiindex(self):
         )
         assert expected == repr(self.mda)
 
-    @pytest.mark.skipif(
-        LooseVersion(np.__version__) < "1.15",
-        reason="old versions of numpy have different printing behavior",
-    )
     def test_repr_multiindex_long(self):
         mindex_long = pd.MultiIndex.from_product(
             [["a", "b", "c", "d"], [1, 2, 3, 4, 5, 6, 7, 8]],