feat(backends): kurtosis

Author: cpcloud

ibis/backends/polars/compiler.py

@@ -792,6 +792,16 @@ def execute_std_var(op, **kw):
     return getattr(arg, method)(ddof=ddof)
 
 
+@translate.register(ops.Kurtosis)
+def execute_kurtosis(op, **kw):
+    arg = translate(op.arg, **kw)
+
+    if (where := op.where) is not None:
+        arg = arg.filter(translate(where, **kw))
+
+    return arg.kurtosis(bias=op.how == "pop")
+
+
 @translate.register(ops.Mode)
 def execute_mode(op, **kw):
     arg = translate(op.arg, **kw)

ibis/backends/sql/compilers/base.py

@@ -7,7 +7,7 @@
 import operator
 import string
 from functools import partial, reduce
-from typing import TYPE_CHECKING, Any, Callable, ClassVar
+from typing import TYPE_CHECKING, Any, Callable, ClassVar, Literal
 
 import sqlglot as sg
 import sqlglot.expressions as sge
@@ -1059,6 +1059,15 @@ def visit_CountDistinctStar(self, op, *, arg, where):
     def visit_CountStar(self, op, *, arg, where):
         return self.agg.count(STAR, where=where)
 
+    def visit_Kurtosis(self, op, *, arg, where, how: Literal["sample", "pop"]):
+        if op.arg.dtype.is_boolean():
+            arg = self.cast(arg, dt.int32)
+
+        if how == "sample":
+            return self.agg.kurtosis(arg, where=where)
+        else:
+            return self.agg.kurtosis_pop(arg, where=where)
+
     def visit_Sum(self, op, *, arg, where):
         if op.arg.dtype.is_boolean():
             arg = self.cast(arg, dt.int32)

ibis/backends/sql/compilers/clickhouse.py

@@ -55,6 +55,7 @@ class ClickHouseCompiler(SQLGlotCompiler):
         ops.StringToDate,
         ops.StringToTime,
         ops.Levenshtein,
+        ops.Kurtosis,
     )
 
     SIMPLE_OPS = {

ibis/backends/sql/compilers/datafusion.py

@@ -44,6 +44,10 @@ class DataFusionCompiler(SQLGlotCompiler):
         ops.StringToDate,
         ops.StringToTimestamp,
         ops.StringToTime,
+        # in theory possible via
+        # https://github.com/datafusion-contrib/datafusion-functions-extra but
+        # not clear how to use that library from Python
+        ops.Kurtosis,
     )
 
     SIMPLE_OPS = {

ibis/backends/sql/compilers/druid.py

@@ -47,6 +47,7 @@ class DruidCompiler(SQLGlotCompiler):
         ops.IntervalFromInteger,
         ops.IsNan,
         ops.IsInf,
+        ops.Kurtosis,
         ops.Levenshtein,
         ops.Median,
         ops.RandomScalar,

ibis/backends/sql/compilers/exasol.py

@@ -58,6 +58,7 @@ class ExasolCompiler(SQLGlotCompiler):
         ops.IntervalFromInteger,
         ops.IsInf,
         ops.IsNan,
+        ops.Kurtosis,
         ops.Levenshtein,
         ops.MultiQuantile,
         ops.RandomUUID,

ibis/backends/sql/compilers/flink.py

@@ -90,6 +90,7 @@ class FlinkCompiler(SQLGlotCompiler):
         ops.RowID,
         ops.Translate,
         ops.StringToTime,
+        ops.Kurtosis,
     )
 
     SIMPLE_OPS = {

ibis/backends/sql/compilers/impala.py

@@ -57,6 +57,7 @@ class ImpalaCompiler(SQLGlotCompiler):
         ops.TimestampBucket,
         ops.TimestampDelta,
         ops.Unnest,
+        ops.Kurtosis,
     )
 
     SIMPLE_OPS = {

ibis/backends/sql/compilers/mssql.py

@@ -100,6 +100,7 @@ class MSSQLCompiler(SQLGlotCompiler):
         ops.Covariance,
         ops.CountDistinctStar,
         ops.DateDiff,
+        ops.Kurtosis,
         ops.IntervalAdd,
         ops.IntervalSubtract,
         ops.IntervalMultiply,

ibis/backends/sql/compilers/mysql.py

@@ -72,6 +72,7 @@ def POS_INF(self):
         ops.ArgMax,
         ops.ArgMin,
         ops.Covariance,
+        ops.Kurtosis,
         ops.Levenshtein,
         ops.Median,
         ops.Mode,