feat: add weighted_average aggregation function support (#208)

## What does this PR do?

add weighted_average aggreagtion function support for
llamastack/llama-stack#1708

Author: SLR722

src/llama_stack_client/lib/cli/eval/run_benchmark.py

@@ -13,7 +13,13 @@
 from tqdm.rich import tqdm
 
 from ..common.utils import create_bar_chart
-from .utils import aggregate_accuracy, aggregate_average, aggregate_categorical_count, aggregate_median
+from .utils import (
+    aggregate_accuracy,
+    aggregate_average,
+    aggregate_weighted_average,
+    aggregate_categorical_count,
+    aggregate_median,
+)
 
 
 @click.command("run-benchmark")
@@ -144,6 +150,8 @@ def run_benchmark(
                             output_res[scoring_fn].append(aggregate_categorical_count(scoring_results))
                         elif aggregation_function == "average":
                             output_res[scoring_fn].append(aggregate_average(scoring_results))
+                        elif aggregation_function == "weighted_average":
+                            output_res[scoring_fn].append(aggregate_weighted_average(scoring_results))
                         elif aggregation_function == "median":
                             output_res[scoring_fn].append(aggregate_median(scoring_results))
                         elif aggregation_function == "accuracy":

src/llama_stack_client/lib/cli/eval/utils.py

@@ -24,6 +24,19 @@ def aggregate_average(
     }
 
 
+def aggregate_weighted_average(
+    scoring_results: List[Dict[str, Union[bool, float, str, List[object], object, None]]],
+) -> Dict[str, Any]:
+    return {
+        "weighted_average": sum(
+            result["score"] * result["weight"]
+            for result in scoring_results
+            if result["score"] is not None and result["weight"] is not None
+        )
+        / sum(result["weight"] for result in scoring_results if result["weight"] is not None),
+    }
+
+
 def aggregate_median(
     scoring_results: List[Dict[str, Union[bool, float, str, List[object], object, None]]],
 ) -> Dict[str, Any]: