AggregationBucket

AggregationBucket represents one bucket returned by Search Aggregation. PyMilvus creates these objects from the server response; applications do not construct them directly.

Python

class pymilvus.AggregationBucket

Properties

• key (list[dict[str, Any]]) -

  The bucket key components. Each item contains field_name, field_id, and value.

• count (int) -

  The number of ANN-retrieved entities assigned to the bucket.

• metrics (dict[str, Any]) -

  Metric values keyed by the aliases defined in SearchAggregation.metrics.

• hits (list[AggregationHit]) -

  Representative entities selected by TopHits. The list is empty when top_hits is not configured at this level.

• sub_groups (list[AggregationBucket]) -

  Nested buckets produced by sub_aggregation. The list is empty at a leaf level.

The collection-valued properties return copies, so changing a returned list or dictionary does not mutate the bucket object.

Example

Python

result = client.search(
    collection_name="products",
    data=[query_vector],
    anns_field="embedding",
    search_aggregation=aggregation,
)

for bucket in result.agg_buckets[0]:
    print(bucket.key, bucket.count, bucket.metrics)
    for hit in bucket.hits:
        print(hit.pk, hit.score, hit.fields)