tfma.metrics.MetricKey

View source on GitHub

A MetricKey uniquely identifies a metric.

tfma.metrics.MetricKey(
    name: str,
    model_name: str = '',
    output_name: str = '',
    sub_key: Optional[tfma.metrics.SubKey] = None,
    aggregation_type: Optional[AggregationType] = None,
    example_weighted: Optional[bool] = False,
    is_diff: Optional[bool] = False
)

Attributes
name	Metric name. Names starting with '' are private and will be filtered from the final results. Names starting with two underscores, '_' are reserved for internal use.
model_name	Optional model name (if multi-model evaluation).
output_name	Optional output name (if multi-output model type).
sub_key	Optional sub key.
aggregation_type	Aggregation type.
example_weighted	Indicates whether this metric was weighted by examples.
is_diff	Optional flag to indicate whether this metrics is a diff metric.

Methods

from_proto

View source

@staticmethod
from_proto(
    pb: metrics_for_slice_pb2.MetricKey
) -> 'MetricKey'

Configures class from proto.

make_baseline_key

View source

make_baseline_key(
    model_name: str
) -> 'MetricKey'

make_diff_key

View source

make_diff_key() -> 'MetricKey'

to_proto

View source

to_proto() -> metrics_for_slice_pb2.MetricKey

Converts key to proto.