tecton.MonitoringConfig

class tecton.MonitoringConfig(monitor_freshness, expected_feature_freshness=None, alert_email=None)

Configuration used to specify monitoring options.

This class describes the FeatureView materialization freshness and alerting configurations. Requires materialization to be enabled. Freshness monitoring requires online materialization to be enabled. See Monitoring Materialization for more details.

Parameters
  • monitor_freshness (bool) – Defines the enabled/disabled state of monitoring when feature data is materialized to the online feature store.

  • expected_feature_freshness (str, optional) – Threshold used to determine if recently materialized feature data is stale. Data is stale if now - anchor_time(most_recent_feature_value) > expected_feature_freshness. Value must be at least 2 times the feature tile length. If not specified, a value determined by the Tecton backend is used

  • alert_email (str, optional) – Email that alerts for this FeatureView will be sent to.

An example declaration of a MonitorConfig

from tecton import batch_feature_view, Input, MonitoringConfig
# For all named arguments to the batch feature view, see docs for details and types.
@batch_feature_view(
    inputs={'credit_scores': Input(credit_scores_batch)},
    # Can be an argument instance to a batch feature view decorator
    monitoring = MonitoringConfig(
        monitor_freshness=True,
        expected_feature_freshness="1w",
        alert_email="jules@tecton.ai"
    ),
    # Other named arguments
    ...
)

# Your batch feature view function
def credit_batch_feature_view(credit_scores):
  ...

Methods

__init__

Initialize self.

__init__(monitor_freshness, expected_feature_freshness=None, alert_email=None)

Initialize self. See help(type(self)) for accurate signature.