TabularObjective

Tabular monitoring objective.

JSON representation 
{
  "featureDriftSpec": {
    object (DataDriftSpec)
  },
  "predictionOutputDriftSpec": {
    object (DataDriftSpec)
  },
  "featureAttributionSpec": {
    object (FeatureAttributionSpec)
  }
}
Fields
featureDriftSpec

object (DataDriftSpec)

Input feature distribution drift monitoring spec.
predictionOutputDriftSpec

object (DataDriftSpec)

Prediction output distribution drift monitoring spec.
featureAttributionSpec

object (FeatureAttributionSpec)

feature attribution monitoring spec.

DataDriftSpec

Data drift monitoring spec. Data drift measures the distribution distance between the current dataset and a baseline dataset. A typical use case is to detect data drift between the recent production serving dataset and the training dataset, or to compare the recent production dataset with a dataset from a previous period.

JSON representation 
{
  "features": [
    string
  ],
  "categoricalMetricType": string,
  "numericMetricType": string,
  "defaultCategoricalAlertCondition": {
    object (ModelMonitoringAlertCondition)
  },
  "defaultNumericAlertCondition": {
    object (ModelMonitoringAlertCondition)
  },
  "featureAlertConditions": {
    string: {
      object (ModelMonitoringAlertCondition)
    },
    ...
  }
}
Fields
features[]

string

feature names / Prediction output names interested in monitoring. These should be a subset of the input feature names or prediction output names specified in the monitoring schema. If the field is not specified all features / prediction outputs outlied in the monitoring schema will be used.
categoricalMetricType

string

Supported metrics type: * l_infinity * jensen_shannon_divergence
numericMetricType

string

Supported metrics type: * jensen_shannon_divergence
defaultCategoricalAlertCondition

object (ModelMonitoringAlertCondition)

Default alert condition for all the categorical features.
defaultNumericAlertCondition

object (ModelMonitoringAlertCondition)

Default alert condition for all the numeric features.
featureAlertConditions

map (key: string, value: object (ModelMonitoringAlertCondition))

Per feature alert condition will override default alert condition.

ModelMonitoringAlertCondition

Monitoring alert triggered condition.

JSON representation 
{

  // Union field condition can be only one of the following:
  "threshold": number
  // End of list of possible types for union field condition.
}
Fields
Union field condition. Alert triggered condition. condition can be only one of the following:
threshold

number

A condition that compares a stats value against a threshold. Alert will be triggered if value above the threshold.

FeatureAttributionSpec

feature attribution monitoring spec.

JSON representation 
{
  "features": [
    string
  ],
  "defaultAlertCondition": {
    object (ModelMonitoringAlertCondition)
  },
  "featureAlertConditions": {
    string: {
      object (ModelMonitoringAlertCondition)
    },
    ...
  },
  "batchExplanationDedicatedResources": {
    object (BatchDedicatedResources)
  }
}
Fields
features[]

string

feature names interested in monitoring. These should be a subset of the input feature names specified in the monitoring schema. If the field is not specified all features outlied in the monitoring schema will be used.
defaultAlertCondition

object (ModelMonitoringAlertCondition)

Default alert condition for all the features.
featureAlertConditions

map (key: string, value: object (ModelMonitoringAlertCondition))

Per feature alert condition will override default alert condition.
batchExplanationDedicatedResources

object (BatchDedicatedResources)

The config of resources used by the Model Monitoring during the batch explanation for non-AutoML models. If not set, n1-standard-2 machine type will be used by default.