Change matplotlib as an optional dependency (#2029)

This PR proposes to change matplotlib as an optional dependency.

```python
>>> from databricks import koalas as ks
>>> ks.range(100).plot.bar()
Traceback (most recent call last):
  ...
ImportError: matplotlib is required for plotting when the default backend 'matplotlib' is selected.
```

Resolves #issues

databricks/koalas/frame.py

@@ -44,7 +44,6 @@
     TYPE_CHECKING,
 )
 
-import matplotlib
 import numpy as np
 import pandas as pd
 from pandas.api.types import is_list_like, is_dict_like, is_scalar
@@ -858,12 +857,12 @@ def add(self, other) -> "DataFrame":
     # create accessor for Koalas specific methods.
     koalas = CachedAccessor("koalas", KoalasFrameMethods)
 
-    def hist(self, bins=10, **kwds) -> matplotlib.axes.Axes:
+    def hist(self, bins=10, **kwds):
         return self.plot.hist(bins, **kwds)
 
     hist.__doc__ = KoalasPlotAccessor.hist.__doc__
 
-    def kde(self, bw_method=None, ind=None, **kwds) -> matplotlib.axes.Axes:
+    def kde(self, bw_method=None, ind=None, **kwds):
         return self.plot.kde(bw_method, ind, **kwds)
 
     kde.__doc__ = KoalasPlotAccessor.kde.__doc__

databricks/koalas/plot/__init__.py

@@ -14,4 +14,3 @@
 # limitations under the License.
 #
 from databricks.koalas.plot.core import *  # noqa: F401
-from databricks.koalas.plot.matplotlib import *  # noqa: F401

databricks/koalas/series.py

@@ -27,7 +27,6 @@
 from itertools import chain
 from typing import Any, Generic, Iterable, List, Optional, Tuple, TypeVar, Union, cast
 
-import matplotlib
 import numpy as np
 import pandas as pd
 from pandas.core.accessor import CachedAccessor
@@ -3004,7 +3003,7 @@ def sample(
 
     sample.__doc__ = DataFrame.sample.__doc__
 
-    def hist(self, bins=10, **kwds) -> matplotlib.axes.Axes:
+    def hist(self, bins=10, **kwds):
         return self.plot.hist(bins, **kwds)
 
     hist.__doc__ = KoalasPlotAccessor.hist.__doc__

docs/source/getting_started/install.rst

@@ -124,7 +124,6 @@ Package       Required version
 `pandas`      >=0.23.2
 `pyspark`     >=2.4.0
 `pyarrow`     >=0.10
-`matplotlib`  >=3.0.0,<3.3.0
 `numpy`       >=1.14
 ============= ================
 
@@ -137,4 +136,5 @@ Package       Required version
 ============= ================
 `mlflow`      >=1.0
 `plotly`      >=4.8
+`matplotlib`  >=3.0.0,<3.3.0
 ============= ================

postBuild

@@ -19,5 +19,5 @@
 
 # Install PySpark manually because it does not exist in requirement file.
 # This file is used in Binder integration.
-pip install 'pyspark>=2.4'
+pip install 'pyspark>=2.4' 'matplotlib>=3.0.0,<3.3.0' 'plotly>=4.8'
 echo "export PYARROW_IGNORE_TIMEZONE=1" >> ~/.profile

requirements-dev.txt

@@ -1,12 +1,12 @@
 # Dependencies in Koalas. When you update don't forget to update setup.py and install.rst in docs.
 pandas>=0.23.2,<1.2.0
 pyarrow>=0.10
-matplotlib>=3.0.0,<3.3.0
 numpy>=1.14,<1.20.0
 
 # Optional dependencies in Koalas.
 mlflow>=1.0
 plotly>=4.8
+matplotlib>=3.0.0,<3.3.0
 
 # Documentation build.
 sphinx>=2.0.0,<3.1.0

setup.py

@@ -51,13 +51,13 @@
         'spark': ['pyspark>=2.4.0'],
         'mlflow': ['mlflow>=1.0'],
         'plotly': ['plotly>=4.8'],
+        'matplotlib': ['matplotlib>=3.0.0,<3.3.0'],
     },
     python_requires='>=3.5,<3.9',
     install_requires=[
         'pandas>=0.23.2,<1.2.0',
         'pyarrow>=0.10',
         'numpy>=1.14,<1.20.0',
-        'matplotlib>=3.0.0,<3.3.0',
     ],
     author="Databricks",
     author_email="[email protected]",