[FEA] Implement .describe() for DataFrameGroupBy

python/cudf/cudf/core/groupby/groupby.py

@@ -12,6 +12,21 @@
 from cudf.utils.utils import GetAttrGetItemMixin, cached_property
 
 
+# The three functions below return the quantiles [25%, 50%, 75%]
+# respectively, which are called in the describe() method to ouput
+# the summary stats of a GroupBy object
+def _quantile_25(x):
+    return x.quantile(0.25)
+
+
+def _quantile_50(x):
+    return x.quantile(0.50)
+
+
+def _quantile_75(x):
+    return x.quantile(0.75)
+
+
 # Note that all valid aggregation methods (e.g. GroupBy.min) are bound to the
 # class after its definition (see below).
 class GroupBy(Serializable):
@@ -597,6 +612,33 @@ def func(x):
 
         return self.agg(func)
 
+    def describe(self):
+        """Return descriptive stats of the values in each group"""
+
+        res = self.agg(
+            [
+                "count",
+                "mean",
+                "std",
+                "min",
+                _quantile_25,
+                _quantile_50,
+                _quantile_75,
+                "max",
+            ]
+        )
+        res.columns = [
+            "count",
+            "mean",
+            "std",
+            "min",
+            "25%",
+            "50%",
+            "75%",
+            "max",
+        ]
+        return res
+
     def sum(self):
         """Compute the column-wise sum of the values in each group."""
         return self.agg("sum")

python/cudf/cudf/tests/test_groupby.py

@@ -1666,3 +1666,17 @@ def test_groupby_mix_agg_scan():
     gb.agg(func[1:])
     with pytest.raises(NotImplementedError, match=err_msg):
         gb.agg(func)
+
+
+@pytest.mark.parametrize(
+    "data", [{"Speed": [380.0, 370.0, 24.0, 26.0], "Score": [50, 30, 90, 80]}],
+)
+@pytest.mark.parametrize("group", ["Score", "Speed"])
+def test_groupby_describe(data, group):
+    pdf = pd.DataFrame(data)
+    gdf = cudf.from_pandas(pdf)
+
+    got = gdf.groupby(group).describe()
+    expect = pdf.groupby(group).describe()
+
+    assert_groupby_results_equal(expect, got, check_dtype=False)