fix(utils): fix possible name collision in train_test_split (ibis-project#142)

Co-authored-by: Deepyaman Datta <deepyaman.datta@utexas.edu>

Author: jitingxu1

ibis_ml/utils/_split.py

@@ -76,20 +76,28 @@ def train_test_split(
         random.seed(random_seed)
 
     # Generate a random 256-bit key
-    random_key = str(random.getrandbits(256))
+    random_str = str(random.getrandbits(256))
 
     if isinstance(unique_key, str):
         unique_key = [unique_key]
 
+    # Append random string to the name to avoid collision
+    combined_key = f"combined_key_{random_str}"
+    train_flag = f"train_{random_str}"
+
     table = table.mutate(
-        combined_key=ibis.literal(",").join(
-            table[col].cast("str") for col in unique_key
-        )
+        **{
+            combined_key: ibis.literal(",").join(
+                table[col].cast("str") for col in unique_key
+            )
+        }
     ).mutate(
-        train=(_.combined_key + random_key).hash().abs() % num_buckets
-        < int((1 - test_size) * num_buckets)
+        **{
+            train_flag: (_[combined_key] + random_str).hash().abs() % num_buckets
+            < int((1 - test_size) * num_buckets)
+        }
     )
 
-    return table[table.train].drop(["combined_key", "train"]), table[~table.train].drop(
-        ["combined_key", "train"]
-    )
+    return table[table[train_flag]].drop([combined_key, train_flag]), table[
+        ~table[train_flag]
+    ].drop([combined_key, train_flag])