diff --git a/machine_learning/apriori_algorithm.py b/machine_learning/apriori_algorithm.py
index 5c3e2baba2c2..560b92f2fb71 100644
--- a/machine_learning/apriori_algorithm.py
+++ b/machine_learning/apriori_algorithm.py
@@ -45,16 +45,16 @@ def prune(itemset: list, candidates: list, length: int) -> list:
     >>> prune(itemset, candidates, 3)
     []
     """
+    # Use a set for O(1) membership and a Counter for multiplicity checks to
+    # preserve robustness for edge cases and match existing doctest behavior.
+    itemset_set = {tuple(item) for item in itemset}
     itemset_counter = Counter(tuple(item) for item in itemset)
     pruned = []
     for candidate in candidates:
         is_subsequence = True
         for item in candidate:
             item_tuple = tuple(item)
-            if (
-                item_tuple not in itemset_counter
-                or itemset_counter[item_tuple] < length - 1
-            ):
+            if item_tuple not in itemset_set or itemset_counter[item_tuple] < length - 1:
                 is_subsequence = False
                 break
         if is_subsequence: