[Docs] Enhance transforming-data documentation with Polars operations section

Signed-off-by: peterxcli <peterxcli@gmail.com>

Author: peterxcli

doc/source/data/transforming-data.rst

@@ -212,7 +212,7 @@ In this case, your function would look like:
         # yield the same batch multiple times
         for _ in range(10):
             yield batch
-            
+
 Choosing the right batch format
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
@@ -249,6 +249,19 @@ program might run into out-of-memory (OOM) errors.
 
 If you encounter an OOM errors, try decreasing your ``batch_size``.
 
+Enabling Polars operations
+~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+You can enable Polars globally to optimize certain Ray Data operations. Ray Data uses Polars internally for better performance when processing data.
+
+To enable Polars operations, configure the :class:`~ray.data.DataContext`:
+
+.. testcode::
+    ctx = ray.data.DataContext.get_current()
+    ctx.use_polars_sort = True
+
+When you enable these flags, Ray Data automatically uses Polars for use Polars for tabular dataset sorting operations. which can significantly improve performance for certain workloads. This doesn't affect your UDF code—you can still use any batch format in :meth:`~ray.data.Dataset.map_batches`.
+
 
 .. _stateful_transforms:
 
@@ -365,7 +378,7 @@ You can read more about resources in Ray here: :ref:`resource-requirements`.
     :hide:
 
     import ray
-    
+
     ds = ray.data.range(1)
 
 .. testcode::