DataRobot API resources > API reference documentation > DataRobot REST API > Monitoring > Data drift

Observability Drift¶

This page outlines the operations, endpoints, parameters, and example requests and responses for the Observability Drift.

GET /api/v2/deployments/{deploymentId}/featureDrift/¶

Retrieve drift scores for features of the deployment.

Code samples¶


curl -X GET https://app.datarobot.com/api/v2/deployments/{deploymentId}/featureDrift/ \
  -H "Accept: application/json" \
  -H "Authorization: Bearer {access-token}"

Parameters

Name	In	Type	Required	Description
start	query	string,null(date-time)	false	Start of the period to retrieve monitoring data, defaults to 7 days ago from the end of the period. Note: this field only accepts top of the hour RFC3339 datetime strings, for example: `2019-08-01T00:00:00Z`.
end	query	string,null(date-time)	false	End of the period to retrieve monitoring data, defaults to the next top of the hour. Note: this field only accepts top of the hour RFC3339 datetime strings, for example: `2019-08-01T00:00:00Z`.
modelId	query	string	false	ID of the model in the deployment. If not set, defaults to the deployment current model.
metric	query	string	false	Name of the metric used to calculate the drift. Can be one of psi, kl_divergence, dissimilarity, hellinger, and js_divergence. Defaults to psi.
offset	query	integer	false	The number of features to skip, defaults to 0.
limit	query	integer	false	The number of features to return, defaults to 25.
segmentAttribute	query	string	false	The name of a segment attribute used for segment analysis.
segmentValue	query	string,null	false	The value of the `segmentAttribute` to segment on.
batchId	query	any	false	The id of the batch for which metrics are being retrieved.
deploymentId	path	string	true	Unique identifier of the deployment.

Enumerated Values¶

Parameter	Value
metric	[`psi`, `kl_divergence`, `dissimilarity`, `hellinger`, `js_divergence`]

Example responses¶

200 Response


{
  "batchId": [],
  "count": 0,
  "data": [
    {
      "baselineSampleSize": 0,
      "driftScore": 0,
      "featureImpact": 0,
      "name": "string",
      "sampleSize": 0,
      "type": "numeric"
    }
  ],
  "metric": "psi",
  "modelId": "string",
  "next": "string",
  "period": {
    "end": "2019-08-24T14:15:22Z",
    "start": "2019-08-24T14:15:22Z"
  },
  "previous": "string",
  "segmentAttribute": "string",
  "segmentValue": "string"
}

Responses¶

Status	Meaning	Description	Schema
200	OK	Features drift over specified time period retrieved.	FeatureDriftResponse

To perform this operation, you must be authenticated by means of one of the following methods:

BearerAuth

GET /api/v2/deployments/{deploymentId}/featureDriftOverBatch/¶

Retrieve drift over batch info for a feature of the deployment.

Code samples¶


curl -X GET https://app.datarobot.com/api/v2/deployments/{deploymentId}/featureDriftOverBatch/?featureNames=string \
  -H "Accept: application/json" \
  -H "Authorization: Bearer {access-token}"

Parameters

Name	In	Type	Required	Description
segmentAttribute	query	string,null	false	The name of the segment on which segment analysis is being performed.
segmentValue	query	string,null	false	The value of the `segmentAttribute` to segment on.
batchId	query	any	false	The id of the batch for which metrics are being retrieved.
featureNames	query	any	true	List of feature names, limited to two per request.
driftMetric	query	string,null	false	The metric used to calculate data drift scores.
modelId	query	string	false	The id of the model for which metrics are being retrieved.
deploymentId	path	string	true	Unique identifier of the deployment.

Enumerated Values¶

Parameter	Value
driftMetric	[`psi`, `kl_divergence`, `dissimilarity`, `hellinger`, `js_divergence`]

Example responses¶

200 Response


{
  "buckets": [
    {
      "baselineSampleSize": 0,
      "batch": {
        "earliestPredictionTimestamp": "2019-08-24T14:15:22Z",
        "id": "string",
        "latestPredictionTimestamp": "2019-08-24T14:15:22Z",
        "name": "string"
      },
      "driftScore": 0,
      "featureImpact": 0,
      "featureName": "string",
      "sampleSize": 0
    }
  ],
  "driftMetric": "psi"
}

Responses¶

Status	Meaning	Description	Schema
200	OK	Drift over batch info for a feature of the deployment.	FeatureDriftOverBatchResponse

To perform this operation, you must be authenticated by means of one of the following methods:

BearerAuth

GET /api/v2/deployments/{deploymentId}/featureDriftOverSpace/¶

Retrieve drift scores for a feature of the deployment over the geospatial feature data.

Notice: Endpoint is currently in [PUBLIC_PREVIEW]. Do not use it in production workflows to reduce risk. See details:

This endpoint depends on the following features that are subject to change.

Feature Flag	Maturity	Enabled by default	Description
MMM_GEOSPATIAL_FEATURES_MONITORING	PUBLIC_PREVIEW	true	Enables monitoring geospatial features.

Code samples¶


curl -X GET https://app.datarobot.com/api/v2/deployments/{deploymentId}/featureDriftOverSpace/ \
  -H "Accept: application/json" \
  -H "Authorization: Bearer {access-token}"

Parameters

Name	In	Type	Required	Description
start	query	string,null(date-time)	false	Start of the period to retrieve monitoring data, defaults to 7 days ago from the end of the period. Note: this field only accepts top of the hour RFC3339 datetime strings, for example: `2019-08-01T00:00:00Z`.
end	query	string,null(date-time)	false	End of the period to retrieve monitoring data, defaults to the next top of the hour. Note: this field only accepts top of the hour RFC3339 datetime strings, for example: `2019-08-01T00:00:00Z`.
geoFeatureName	query	string	false	The name of the geospatial feature. Segmented analysis must be enabled for the feature specified.
featureName	query	string	false	The name of the feature to retrieve drift scores for. Defaults to the target or, if there is no target, the most important feature.
modelId	query	string	false	The ID of the model that feature drift is being retrieved from.
metric	query	string	false	The metric used to calculate drift score.
deploymentId	path	string	true	Unique identifier of the deployment.

Enumerated Values¶

Parameter	Value
metric	[`psi`, `klDivergence`, `dissimilarity`, `hellinger`, `jsDivergence`]

Example responses¶

200 Response


{
  "data": [
    {
      "baselineSampleSize": 0,
      "driftScore": 0,
      "hexagon": "string",
      "sampleSize": 0
    }
  ],
  "featureName": "string",
  "geoFeatureName": "string",
  "period": {
    "end": "2019-08-24T14:15:22Z",
    "start": "2019-08-24T14:15:22Z"
  }
}

Responses¶

Status	Meaning	Description	Schema
200	OK	The retrieved drift scores for the specified feature over the specified geospatial feature data.	FeatureDriftOverSpaceResponse

To perform this operation, you must be authenticated by means of one of the following methods:

BearerAuth

GET /api/v2/deployments/{deploymentId}/featureDriftOverTime/¶

Retrieve drift over time info for a feature of the deployment.

Code samples¶


curl -X GET https://app.datarobot.com/api/v2/deployments/{deploymentId}/featureDriftOverTime/?featureNames=string \
  -H "Accept: application/json" \
  -H "Authorization: Bearer {access-token}"

Parameters

Name	In	Type	Required	Description
start	query	string,null(date-time)	false	Start of the period to retrieve monitoring data, defaults to 7 days ago from the end of the period. Note: this field only accepts top of the hour RFC3339 datetime strings, for example: `2019-08-01T00:00:00Z`.
end	query	string,null(date-time)	false	End of the period to retrieve monitoring data, defaults to the next top of the hour. Note: this field only accepts top of the hour RFC3339 datetime strings, for example: `2019-08-01T00:00:00Z`.
bucketSize	query	string(duration)	false	The time duration of a bucket. Needs to be multiple of one hour. Can not be longer than the total length of the period. If not set, a default value will be calculated based on the start and end time.
modelId	query	string	false	The id of the model for which the features drift is being retrieved.
featureNames	query	any	true	List of feature names, limited to two per request.
metric	query	string	false	Name of the metric used to calculate the drift. Can be one of psi, kl_divergence, dissimilarity, hellinger, and js_divergence. Defaults to psi.
segmentAttribute	query	string,null	false	The name of the segment on which segment analysis is being performed.
segmentValue	query	string,null	false	The value of the `segmentAttribute` to segment on.
deploymentId	path	string	true	Unique identifier of the deployment.

Enumerated Values¶

Parameter	Value
metric	[`psi`, `kl_divergence`, `dissimilarity`, `hellinger`, `js_divergence`]

Example responses¶

200 Response


{
  "buckets": [
    {
      "baselineSampleSize": 0,
      "driftScore": 0,
      "featureImpact": 0,
      "featureName": "string",
      "period": {
        "end": "2019-08-24T14:15:22Z",
        "start": "2019-08-24T14:15:22Z"
      },
      "sampleSize": 0
    }
  ],
  "features": [
    {
      "featureName": "string",
      "featureType": "numeric"
    }
  ],
  "metric": "string",
  "summaries": [
    {
      "featureImpact": 0,
      "featureName": "string"
    }
  ]
}

Responses¶

Status	Meaning	Description	Schema
200	OK	Feature drift over time info of the deployment retrieved.	FeatureDriftOverTimeResponse
400	Bad Request	Request invalid, refer to messages for detail.	None
403	Forbidden	Model Deployments and/or Monitoring are not enabled.	None
404	Not Found	Either the deployment does not exist or user does not have permission to view the deployment.	None

To perform this operation, you must be authenticated by means of one of the following methods:

BearerAuth

GET /api/v2/deployments/{deploymentId}/predictionsOverBatch/¶

Retrieve metrics about predictions, such as row count, mean predicted value, mean probabilities, and class distribution, over batches.

Code samples¶


curl -X GET https://app.datarobot.com/api/v2/deployments/{deploymentId}/predictionsOverBatch/ \
  -H "Accept: application/json" \
  -H "Authorization: Bearer {access-token}"

Parameters

Name	In	Type	Required	Description
segmentAttribute	query	string,null	false	The name of the segment on which segment analysis is being performed.
segmentValue	query	string,null	false	The value of the `segmentAttribute` to segment on.
batchId	query	any	false	The id of the batch for which metrics are being retrieved.
modelId	query	string	false	The id of the model for which metrics are being retrieved.
includePercentiles	query	string	false	Include percentiles in the response, only applicable to deployments with binary classification, location and regression target.
deploymentId	path	string	true	Unique identifier of the deployment.

Enumerated Values¶

Parameter	Value
includePercentiles	[`false`, `False`, `true`, `True`]

Example responses¶

200 Response


{
  "baselines": [
    {
      "classDistribution": [
        {
          "className": "string",
          "count": 0,
          "percent": 0
        }
      ],
      "meanGeoCentroid": {
        "latitude": 0,
        "longitude": 0
      },
      "meanPredictedValue": 0,
      "meanProbabilities": [
        {
          "className": "string",
          "value": 0
        }
      ],
      "modelId": "string",
      "percentiles": [
        {
          "geoCentroid": {
            "latitude": 0,
            "longitude": 0
          },
          "percent": 0,
          "value": 0
        }
      ],
      "predictionsWarningCount": 0,
      "rowCount": 0
    }
  ],
  "buckets": [
    {
      "batch": {
        "earliestPredictionTimestamp": "2019-08-24T14:15:22Z",
        "id": "string",
        "latestPredictionTimestamp": "2019-08-24T14:15:22Z",
        "name": "string"
      },
      "classDistribution": [
        {
          "className": "string",
          "count": 0,
          "percent": 0
        }
      ],
      "meanGeoCentroid": {
        "latitude": 0,
        "longitude": 0
      },
      "meanPredictedValue": 0,
      "meanProbabilities": [
        {
          "className": "string",
          "value": 0
        }
      ],
      "percentiles": [
        {
          "geoCentroid": {
            "latitude": 0,
            "longitude": 0
          },
          "percent": 0,
          "value": 0
        }
      ],
      "predictionsWarningCount": 0,
      "rowCount": 0
    }
  ],
  "segmentAttribute": "string",
  "segmentValue": ""
}

Responses¶

Status	Meaning	Description	Schema
200	OK	Predictions over batch info retrieved.	PredictionsOverBatchResponse

To perform this operation, you must be authenticated by means of one of the following methods:

BearerAuth

GET /api/v2/deployments/{deploymentId}/predictionsOverSpace/¶

Retrieve predictions stats for the deployment over the geospatial feature data.

Notice: Endpoint is currently in [PUBLIC_PREVIEW]. Do not use it in production workflows to reduce risk. See details:

This endpoint depends on the following features that are subject to change.

Feature Flag	Maturity	Enabled by default	Description
MMM_GEOSPATIAL_FEATURES_MONITORING	PUBLIC_PREVIEW	true	Enables monitoring geospatial features.

Code samples¶


curl -X GET https://app.datarobot.com/api/v2/deployments/{deploymentId}/predictionsOverSpace/ \
  -H "Accept: application/json" \
  -H "Authorization: Bearer {access-token}"

Parameters

Name	In	Type	Required	Description
start	query	string,null(date-time)	false	Start of the period to retrieve monitoring data, defaults to 7 days ago from the end of the period. Note: this field only accepts top of the hour RFC3339 datetime strings, for example: `2019-08-01T00:00:00Z`.
end	query	string,null(date-time)	false	End of the period to retrieve monitoring data, defaults to the next top of the hour. Note: this field only accepts top of the hour RFC3339 datetime strings, for example: `2019-08-01T00:00:00Z`.
geoFeatureName	query	string	false	The name of the geospatial feature. Segmented analysis must be enabled for the feature specified.
modelId	query	string	false	The ID of the model that feature drift is being retrieved from.
targetClass	query	any	false	Target class to filter out results.
includePercentiles	query	string	false	Include percentiles in the response, only applicable to deployments with binary classification, location and regression target.
deploymentId	path	string	true	Unique identifier of the deployment.

Enumerated Values¶

Parameter	Value
includePercentiles	[`false`, `False`, `true`, `True`]

Example responses¶

200 Response


{
  "baselines": [
    {
      "classDistribution": [
        {
          "className": "string",
          "count": 0,
          "percent": 0
        }
      ],
      "hexagon": "string",
      "meanGeoCentroid": {
        "latitude": 0,
        "longitude": 0
      },
      "meanPredictedValue": 0,
      "meanProbabilities": [
        {
          "className": "string",
          "value": 0
        }
      ],
      "percentiles": [
        {
          "geoCentroid": {
            "latitude": 0,
            "longitude": 0
          },
          "percent": 0,
          "value": 0
        }
      ],
      "predictionsWarningCount": 0,
      "rowCount": 0
    }
  ],
  "buckets": [
    {
      "classDistribution": [
        {
          "className": "string",
          "count": 0,
          "percent": 0
        }
      ],
      "hexagon": "string",
      "meanGeoCentroid": {
        "latitude": 0,
        "longitude": 0
      },
      "meanPredictedValue": 0,
      "meanProbabilities": [
        {
          "className": "string",
          "value": 0
        }
      ],
      "percentiles": [
        {
          "geoCentroid": {
            "latitude": 0,
            "longitude": 0
          },
          "percent": 0,
          "value": 0
        }
      ],
      "predictionsWarningCount": 0,
      "rowCount": 0
    }
  ],
  "geoFeatureName": "string",
  "modelId": "string",
  "period": {
    "end": "2019-08-24T14:15:22Z",
    "start": "2019-08-24T14:15:22Z"
  }
}

Responses¶

Status	Meaning	Description	Schema
200	OK	The retrieved prediction stats for the specified feature over the specified geospatial feature data.	PredictionsOverSpaceResponse

To perform this operation, you must be authenticated by means of one of the following methods:

BearerAuth

GET /api/v2/deployments/{deploymentId}/predictionsOverTime/¶

Retrieve metrics about predictions, such as row count, mean predicted value, mean probabilities, and class distribution, over a specific time range.

Code samples¶


curl -X GET https://app.datarobot.com/api/v2/deployments/{deploymentId}/predictionsOverTime/ \
  -H "Accept: application/json" \
  -H "Authorization: Bearer {access-token}"

Parameters

Name	In	Type	Required	Description
start	query	string,null(date-time)	false	Start of the period to retrieve monitoring data, defaults to 7 days ago from the end of the period. Note: this field only accepts top of the hour RFC3339 datetime strings, for example: `2019-08-01T00:00:00Z`.
end	query	string,null(date-time)	false	End of the period to retrieve monitoring data, defaults to the next top of the hour. Note: this field only accepts top of the hour RFC3339 datetime strings, for example: `2019-08-01T00:00:00Z`.
bucketSize	query	string	false	Time duration of buckets
segmentAttribute	query	string,null	false	The name of the segment on which segment analysis is being performed.
segmentValue	query	string,null	false	The value of the `segmentAttribute` to segment on.
modelId	query	any	false	The ID of the models for which metrics are being retrieved.
targetClass	query	any	false	Target class to filter out results.
includePercentiles	query	string	false	Include percentiles in the response, only applicable to deployments with binary classification, location and regression target.
deploymentId	path	string	true	Unique identifier of the deployment.

Enumerated Values¶

Parameter	Value
bucketSize	[`PT1H`, `P1D`, `P7D`, `P1M`]
includePercentiles	[`false`, `False`, `true`, `True`]

Example responses¶

200 Response


{
  "baselines": [
    {
      "classDistribution": [
        {
          "className": "string",
          "count": 0,
          "percent": 0
        }
      ],
      "meanGeoCentroid": {
        "latitude": 0,
        "longitude": 0
      },
      "meanPredictedValue": 0,
      "meanProbabilities": [
        {
          "className": "string",
          "value": 0
        }
      ],
      "modelId": "string",
      "percentiles": [
        {
          "geoCentroid": {
            "latitude": 0,
            "longitude": 0
          },
          "percent": 0,
          "value": 0
        }
      ],
      "predictionsWarningCount": 0,
      "rowCount": 0
    }
  ],
  "buckets": [
    {
      "classDistribution": [
        {
          "className": "string",
          "count": 0,
          "percent": 0
        }
      ],
      "meanGeoCentroid": {
        "latitude": 0,
        "longitude": 0
      },
      "meanPredictedValue": 0,
      "meanProbabilities": [
        {
          "className": "string",
          "value": 0
        }
      ],
      "modelId": "string",
      "percentiles": [
        {
          "geoCentroid": {
            "latitude": 0,
            "longitude": 0
          },
          "percent": 0,
          "value": 0
        }
      ],
      "period": {
        "end": "2019-08-24T14:15:22Z",
        "start": "2019-08-24T14:15:22Z"
      },
      "predictionsWarningCount": 0,
      "rowCount": 0
    }
  ],
  "segmentAttribute": "string",
  "segmentValue": ""
}

Responses¶

Status	Meaning	Description	Schema
200	OK	Predictions over time info retrieved.	PredictionsOverTimeResponse

To perform this operation, you must be authenticated by means of one of the following methods:

BearerAuth

GET /api/v2/deployments/{deploymentId}/targetDrift/¶

Retrieve target drift data.

Code samples¶


curl -X GET https://app.datarobot.com/api/v2/deployments/{deploymentId}/targetDrift/ \
  -H "Accept: application/json" \
  -H "Authorization: Bearer {access-token}"

Parameters

Name	In	Type	Required	Description
start	query	string,null(date-time)	false	Start of the period to retrieve monitoring data, defaults to 7 days ago from the end of the period. Note: this field only accepts top of the hour RFC3339 datetime strings, for example: `2019-08-01T00:00:00Z`.
end	query	string,null(date-time)	false	End of the period to retrieve monitoring data, defaults to the next top of the hour. Note: this field only accepts top of the hour RFC3339 datetime strings, for example: `2019-08-01T00:00:00Z`.
modelId	query	string	false	An ID of the model in the deployment. If not set, defaults to the deployment current model.
metric	query	string	false	Metric used to calculate drift score.
segmentAttribute	query	string	false	The name of a segment attribute used for segment analysis.
segmentValue	query	string,null	false	The value of the `segmentAttribute` to segment on.
batchId	query	any	false	The id of the batch for which metrics are being retrieved.
deploymentId	path	string	true	Unique identifier of the deployment.

Enumerated Values¶

Parameter	Value
metric	[`psi`, `kl_divergence`, `dissimilarity`, `hellinger`, `js_divergence`]

Example responses¶

200 Response


{
  "baselineSampleSize": 0,
  "batchId": [],
  "driftScore": 0,
  "metric": "psi",
  "modelId": "string",
  "period": {
    "end": "2019-08-24T14:15:22Z",
    "start": "2019-08-24T14:15:22Z"
  },
  "sampleSize": 0,
  "segmentAttribute": "string",
  "segmentValue": "",
  "targetName": "string",
  "type": "numeric"
}

Responses¶

Status	Meaning	Description	Schema
200	OK	Target drift over specified time period retrieved.	TargetDriftResponse

To perform this operation, you must be authenticated by means of one of the following methods:

BearerAuth

Schemas¶

ClassDistribution


{
  "className": "string",
  "count": 0,
  "percent": 0
}

Properties¶

Name	Type	Required	Description
className	string	true	Name of the class
count	integer	true	Count of rows labeled with a class in the bucket
percent	number	true	Percent of rows labeled with a class in the bucket

DriftBatch


{
  "earliestPredictionTimestamp": "2019-08-24T14:15:22Z",
  "id": "string",
  "latestPredictionTimestamp": "2019-08-24T14:15:22Z",
  "name": "string"
}

Info of the batch associated with the bucket.

Properties¶

Name	Type	Required	Description
earliestPredictionTimestamp	string(date-time)	true	Earliest prediction timestamp of a batch.
id	string	true	Batch ID.
latestPredictionTimestamp	string,null(date-time)	true	Latest prediction timestamp of a batch.
name	string	true	Batch name.

FeatureDrift


{
  "baselineSampleSize": 0,
  "driftScore": 0,
  "featureImpact": 0,
  "name": "string",
  "sampleSize": 0,
  "type": "numeric"
}

Properties¶

Name	Type	Required	Description
baselineSampleSize	integer	true	The sample size of the training data.
driftScore	number,null	true	The drift score for this feature.
featureImpact	number,null	true	The feature impact score for this feature.
name	string	true	The name of the feature.
sampleSize	integer	true	The number of predictions used to compute the drift score.
type	string	true	Type of the feature.

Enumerated Values¶

Property	Value
type	[`numeric`, `categorical`, `text`]

FeatureDriftOverBatchBucket


{
  "baselineSampleSize": 0,
  "batch": {
    "earliestPredictionTimestamp": "2019-08-24T14:15:22Z",
    "id": "string",
    "latestPredictionTimestamp": "2019-08-24T14:15:22Z",
    "name": "string"
  },
  "driftScore": 0,
  "featureImpact": 0,
  "featureName": "string",
  "sampleSize": 0
}

Properties¶

Name	Type	Required	Description
baselineSampleSize	integer	true	The sample size in the baseline used to calculate drift score.
batch	DriftBatch	true	Info of the batch associated with the bucket.
driftScore	number,null	true	Drift score of the feature.
featureImpact	number	true	The feature impact score of the feature.
featureName	string	true	Feature name.
sampleSize	integer	true	The sample size in the batch used to calculate drift score.

FeatureDriftOverBatchResponse


{
  "buckets": [
    {
      "baselineSampleSize": 0,
      "batch": {
        "earliestPredictionTimestamp": "2019-08-24T14:15:22Z",
        "id": "string",
        "latestPredictionTimestamp": "2019-08-24T14:15:22Z",
        "name": "string"
      },
      "driftScore": 0,
      "featureImpact": 0,
      "featureName": "string",
      "sampleSize": 0
    }
  ],
  "driftMetric": "psi"
}

Properties¶

Name	Type	Required	Restrictions	Description
buckets	[FeatureDriftOverBatchBucket]	true		A list of buckets to display feature drift over batch.
driftMetric	string	false		The metric used to calculate data drift scores.

Enumerated Values¶

Property	Value
driftMetric	[`psi`, `kl_divergence`, `dissimilarity`, `hellinger`, `js_divergence`]

FeatureDriftOverSpaceBucket


{
  "baselineSampleSize": 0,
  "driftScore": 0,
  "hexagon": "string",
  "sampleSize": 0
}

Properties¶

Name	Type	Required	Description
baselineSampleSize	integer	true	The baseline sample size for the hexagon.
driftScore	number	true	The drift score for this feature.
hexagon	string	true	h3 hexagon.
sampleSize	integer	true	The sample size for the hexagon.

FeatureDriftOverSpaceResponse


{
  "data": [
    {
      "baselineSampleSize": 0,
      "driftScore": 0,
      "hexagon": "string",
      "sampleSize": 0
    }
  ],
  "featureName": "string",
  "geoFeatureName": "string",
  "period": {
    "end": "2019-08-24T14:15:22Z",
    "start": "2019-08-24T14:15:22Z"
  }
}

Properties¶

Name	Type	Required	Restrictions	Description
data	[FeatureDriftOverSpaceBucket]	true	maxItems: 10000	An array [DriftObject], each in the form described below
featureName	string	true		The name of the feature to retrieve drift scores for. Defaults to the target or, if there is no target, the most important feature.
geoFeatureName	string	true		The name of the geospatial feature. Segmented analysis must be enabled for the feature specified.
period	TimeRange	false		An object with the keys "start" and "end" defining the period.

FeatureDriftOverTimeBucket


{
  "baselineSampleSize": 0,
  "driftScore": 0,
  "featureImpact": 0,
  "featureName": "string",
  "period": {
    "end": "2019-08-24T14:15:22Z",
    "start": "2019-08-24T14:15:22Z"
  },
  "sampleSize": 0
}

Properties¶

Name	Type	Required	Description
baselineSampleSize	integer	true	The sample size of the training data used during model creation
driftScore	number,null	true	The aggregated drift score for the target.
featureImpact	number	true	The feature impact score of the feature.
featureName	string	true	Name of the feature.
period	TimeRange	true	An object with the keys "start" and "end" defining the period.
sampleSize	integer	true	The sample size in the query period used to calculate drift score.

FeatureDriftOverTimeFeature


{
  "featureName": "string",
  "featureType": "numeric"
}

Properties¶

Name	Type	Required	Restrictions	Description
featureName	string	true		Name of the requested feature.
featureType	string	true		Data type of the requested feature.

Enumerated Values¶

Property	Value
featureType	[`numeric`, `categorical`, `text`]

FeatureDriftOverTimeResponse


{
  "buckets": [
    {
      "baselineSampleSize": 0,
      "driftScore": 0,
      "featureImpact": 0,
      "featureName": "string",
      "period": {
        "end": "2019-08-24T14:15:22Z",
        "start": "2019-08-24T14:15:22Z"
      },
      "sampleSize": 0
    }
  ],
  "features": [
    {
      "featureName": "string",
      "featureType": "numeric"
    }
  ],
  "metric": "string",
  "summaries": [
    {
      "featureImpact": 0,
      "featureName": "string"
    }
  ]
}

Properties¶

Name	Type	Required	Description
buckets	[FeatureDriftOverTimeBucket]	true	A list of aggregated drift scores by feature over a given period.
features	[FeatureDriftOverTimeFeature]	true	A list of the requested features and their feature type.
metric	string	true	Name of requested metric.
summaries	[FeatureDriftOverTimeSummary]	true	A list of aggregated drift scores by feature over a given period.

FeatureDriftOverTimeSummary


{
  "featureImpact": 0,
  "featureName": "string"
}

Properties¶

Name	Type	Required	Restrictions	Description
featureImpact	number	true		The feature impact score of the feature.
featureName	string	true		Name of the feature.

FeatureDriftResponse


{
  "batchId": [],
  "count": 0,
  "data": [
    {
      "baselineSampleSize": 0,
      "driftScore": 0,
      "featureImpact": 0,
      "name": "string",
      "sampleSize": 0,
      "type": "numeric"
    }
  ],
  "metric": "psi",
  "modelId": "string",
  "next": "string",
  "period": {
    "end": "2019-08-24T14:15:22Z",
    "start": "2019-08-24T14:15:22Z"
  },
  "previous": "string",
  "segmentAttribute": "string",
  "segmentValue": "string"
}

Properties¶

Name	Type	Required	Restrictions	Description
batchId	any	false		The id of the batch for which metrics are being retrieved.

oneOf

Name	Type	Required	Restrictions	Description
» anonymous	string	false		none

xor

Name	Type	Required	Restrictions	Description
» anonymous	[string]	false	maxItems: 25	none

continued

Name	Type	Required	Description
count	integer	false	The number of items returned on this page.
data	[FeatureDrift]	true	An array [DriftObject], each in the form described below
metric	string	false	Metric used to calculate drift score.
modelId	string	true	The id of the model for which the features drift is being retrieved.
next	string,null	false	A URL pointing to the next page (if null, there is no next page)
period	TimeRange	false	An object with the keys "start" and "end" defining the period.
previous	string,null	false	A URL pointing to the previous page (if null, there is no previous page)
segmentAttribute	string,null	false	The name of the segment on which segment analysis is being performed.
segmentValue	string,null	false	The value of the `segmentAttribute` to segment on.

Enumerated Values¶

Property	Value
metric	[`psi`, `kl_divergence`, `dissimilarity`, `hellinger`, `js_divergence`]

GeoPoint


{
  "latitude": 0,
  "longitude": 0
}

Geo centroid.

Properties¶

Name	Type	Required	Restrictions	Description
latitude	number	true		Latitude.
longitude	number	true		Longitude.

MeanProbability


{
  "className": "string",
  "value": 0
}

Properties¶

Name	Type	Required	Restrictions	Description
className	string	true		Name of the class
value	number	true		Mean predicted probability for a class for all rows in the bucket

Percentile


{
  "geoCentroid": {
    "latitude": 0,
    "longitude": 0
  },
  "percent": 0,
  "value": 0
}

Properties¶

Name	Type	Required	Description
geoCentroid	GeoPoint	false	Geo centroid.
percent	number	true	Percent of the percentile
value	number,null	false	Predicted value or probability at a percentile

PredictionsOverBatchBucket


{
  "batch": {
    "earliestPredictionTimestamp": "2019-08-24T14:15:22Z",
    "id": "string",
    "latestPredictionTimestamp": "2019-08-24T14:15:22Z",
    "name": "string"
  },
  "classDistribution": [
    {
      "className": "string",
      "count": 0,
      "percent": 0
    }
  ],
  "meanGeoCentroid": {
    "latitude": 0,
    "longitude": 0
  },
  "meanPredictedValue": 0,
  "meanProbabilities": [
    {
      "className": "string",
      "value": 0
    }
  ],
  "percentiles": [
    {
      "geoCentroid": {
        "latitude": 0,
        "longitude": 0
      },
      "percent": 0,
      "value": 0
    }
  ],
  "predictionsWarningCount": 0,
  "rowCount": 0
}

Properties¶

Name	Type	Required	Restrictions	Description
batch	DriftBatch	true		Info of the batch associated with the bucket.
classDistribution	[ClassDistribution]	false	maxItems: 10000	Class distribution for all classes in the bucket, only for classification deployments.
meanGeoCentroid	GeoPoint	false		Geo centroid.
meanPredictedValue	number,null	false		Mean predicted value for all rows in the bucket, only for regression deployments.
meanProbabilities	[MeanProbability]	false	maxItems: 10000	Mean predicted probabilities for all classes in the bucket, only for classification deployments
percentiles	[Percentile]	false	maxItems: 10	Predicted value or positive class predicted probability at specific percentiles in the bucket, only for regression and binary classification deployments.
predictionsWarningCount	integer,null	false		The number of predictions with warning in the bucket
rowCount	integer,null	true		Number of rows in the bucket.

PredictionsOverBatchResponse


{
  "baselines": [
    {
      "classDistribution": [
        {
          "className": "string",
          "count": 0,
          "percent": 0
        }
      ],
      "meanGeoCentroid": {
        "latitude": 0,
        "longitude": 0
      },
      "meanPredictedValue": 0,
      "meanProbabilities": [
        {
          "className": "string",
          "value": 0
        }
      ],
      "modelId": "string",
      "percentiles": [
        {
          "geoCentroid": {
            "latitude": 0,
            "longitude": 0
          },
          "percent": 0,
          "value": 0
        }
      ],
      "predictionsWarningCount": 0,
      "rowCount": 0
    }
  ],
  "buckets": [
    {
      "batch": {
        "earliestPredictionTimestamp": "2019-08-24T14:15:22Z",
        "id": "string",
        "latestPredictionTimestamp": "2019-08-24T14:15:22Z",
        "name": "string"
      },
      "classDistribution": [
        {
          "className": "string",
          "count": 0,
          "percent": 0
        }
      ],
      "meanGeoCentroid": {
        "latitude": 0,
        "longitude": 0
      },
      "meanPredictedValue": 0,
      "meanProbabilities": [
        {
          "className": "string",
          "value": 0
        }
      ],
      "percentiles": [
        {
          "geoCentroid": {
            "latitude": 0,
            "longitude": 0
          },
          "percent": 0,
          "value": 0
        }
      ],
      "predictionsWarningCount": 0,
      "rowCount": 0
    }
  ],
  "segmentAttribute": "string",
  "segmentValue": ""
}

Properties¶

Name	Type	Required	Description
baselines	[TargetBaseline]	true	Target baselines
buckets	[PredictionsOverBatchBucket]	true	Predictions over batch buckets
segmentAttribute	string,null	false	The name of the segment on which segment analysis is being performed.
segmentValue	string,null	false	The value of the `segmentAttribute` to segment on.

PredictionsOverSpaceBucket


{
  "classDistribution": [
    {
      "className": "string",
      "count": 0,
      "percent": 0
    }
  ],
  "hexagon": "string",
  "meanGeoCentroid": {
    "latitude": 0,
    "longitude": 0
  },
  "meanPredictedValue": 0,
  "meanProbabilities": [
    {
      "className": "string",
      "value": 0
    }
  ],
  "percentiles": [
    {
      "geoCentroid": {
        "latitude": 0,
        "longitude": 0
      },
      "percent": 0,
      "value": 0
    }
  ],
  "predictionsWarningCount": 0,
  "rowCount": 0
}

Properties¶

Name	Type	Required	Restrictions	Description
classDistribution	[ClassDistribution]	false	maxItems: 10000	Class distribution for all classes in the bucket, only for classification deployments.
hexagon	string,null	false		h3 hexagon.
meanGeoCentroid	GeoPoint	false		Geo centroid.
meanPredictedValue	number,null	false		Mean predicted value for all rows in the bucket, only for regression deployments.
meanProbabilities	[MeanProbability]	false	maxItems: 10000	Mean predicted probabilities for all classes in the bucket, only for classification deployments
percentiles	[Percentile]	false	maxItems: 10	Predicted value or positive class predicted probability at specific percentiles in the bucket, only for regression and binary classification deployments.
predictionsWarningCount	integer,null	false		The number of predictions with warning in the bucket
rowCount	integer,null	true		Number of rows in the bucket.

PredictionsOverSpaceResponse


{
  "baselines": [
    {
      "classDistribution": [
        {
          "className": "string",
          "count": 0,
          "percent": 0
        }
      ],
      "hexagon": "string",
      "meanGeoCentroid": {
        "latitude": 0,
        "longitude": 0
      },
      "meanPredictedValue": 0,
      "meanProbabilities": [
        {
          "className": "string",
          "value": 0
        }
      ],
      "percentiles": [
        {
          "geoCentroid": {
            "latitude": 0,
            "longitude": 0
          },
          "percent": 0,
          "value": 0
        }
      ],
      "predictionsWarningCount": 0,
      "rowCount": 0
    }
  ],
  "buckets": [
    {
      "classDistribution": [
        {
          "className": "string",
          "count": 0,
          "percent": 0
        }
      ],
      "hexagon": "string",
      "meanGeoCentroid": {
        "latitude": 0,
        "longitude": 0
      },
      "meanPredictedValue": 0,
      "meanProbabilities": [
        {
          "className": "string",
          "value": 0
        }
      ],
      "percentiles": [
        {
          "geoCentroid": {
            "latitude": 0,
            "longitude": 0
          },
          "percent": 0,
          "value": 0
        }
      ],
      "predictionsWarningCount": 0,
      "rowCount": 0
    }
  ],
  "geoFeatureName": "string",
  "modelId": "string",
  "period": {
    "end": "2019-08-24T14:15:22Z",
    "start": "2019-08-24T14:15:22Z"
  }
}

Properties¶

Name	Type	Required	Restrictions	Description
baselines	[PredictionsOverSpaceBucket]	true	maxItems: 10000	Baseline predictions per geospatial hexagon.
buckets	[PredictionsOverSpaceBucket]	true	maxItems: 10000	Predictions per geospatial hexagon.
geoFeatureName	string	true		The name of the geospatial feature. Segmented analysis must be enabled for the feature specified.
modelId	string	false		The ID of the model that predictions are being retrieved from.
period	TimeRange	false		An object with the keys "start" and "end" defining the period.

PredictionsOverTimeBucket


{
  "classDistribution": [
    {
      "className": "string",
      "count": 0,
      "percent": 0
    }
  ],
  "meanGeoCentroid": {
    "latitude": 0,
    "longitude": 0
  },
  "meanPredictedValue": 0,
  "meanProbabilities": [
    {
      "className": "string",
      "value": 0
    }
  ],
  "modelId": "string",
  "percentiles": [
    {
      "geoCentroid": {
        "latitude": 0,
        "longitude": 0
      },
      "percent": 0,
      "value": 0
    }
  ],
  "period": {
    "end": "2019-08-24T14:15:22Z",
    "start": "2019-08-24T14:15:22Z"
  },
  "predictionsWarningCount": 0,
  "rowCount": 0
}

Properties¶

Name	Type	Required	Restrictions	Description
classDistribution	[ClassDistribution]	false	maxItems: 10000	Class distribution for all classes in the bucket, only for classification deployments.
meanGeoCentroid	GeoPoint	false		Geo centroid.
meanPredictedValue	number,null	false		Mean predicted value for all rows in the bucket, only for regression deployments.
meanProbabilities	[MeanProbability]	false	maxItems: 10000	Mean predicted probabilities for all classes in the bucket, only for classification deployments
modelId	string	true		ID of the model
percentiles	[Percentile]	false	maxItems: 10	Predicted value or positive class predicted probability at specific percentiles in the bucket, only for regression and binary classification deployments.
period	TimeRange	true		An object with the keys "start" and "end" defining the period.
predictionsWarningCount	integer,null	false		The number of predictions with warning in the bucket
rowCount	integer,null	true		Number of rows in the bucket.

PredictionsOverTimeResponse


{
  "baselines": [
    {
      "classDistribution": [
        {
          "className": "string",
          "count": 0,
          "percent": 0
        }
      ],
      "meanGeoCentroid": {
        "latitude": 0,
        "longitude": 0
      },
      "meanPredictedValue": 0,
      "meanProbabilities": [
        {
          "className": "string",
          "value": 0
        }
      ],
      "modelId": "string",
      "percentiles": [
        {
          "geoCentroid": {
            "latitude": 0,
            "longitude": 0
          },
          "percent": 0,
          "value": 0
        }
      ],
      "predictionsWarningCount": 0,
      "rowCount": 0
    }
  ],
  "buckets": [
    {
      "classDistribution": [
        {
          "className": "string",
          "count": 0,
          "percent": 0
        }
      ],
      "meanGeoCentroid": {
        "latitude": 0,
        "longitude": 0
      },
      "meanPredictedValue": 0,
      "meanProbabilities": [
        {
          "className": "string",
          "value": 0
        }
      ],
      "modelId": "string",
      "percentiles": [
        {
          "geoCentroid": {
            "latitude": 0,
            "longitude": 0
          },
          "percent": 0,
          "value": 0
        }
      ],
      "period": {
        "end": "2019-08-24T14:15:22Z",
        "start": "2019-08-24T14:15:22Z"
      },
      "predictionsWarningCount": 0,
      "rowCount": 0
    }
  ],
  "segmentAttribute": "string",
  "segmentValue": ""
}

Properties¶

Name	Type	Required	Description
baselines	[TargetBaseline]	true	Target baselines
buckets	[PredictionsOverTimeBucket]	true	Predictions over time buckets
segmentAttribute	string,null	false	The name of the segment on which segment analysis is being performed.
segmentValue	string,null	false	The value of the `segmentAttribute` to segment on.

TargetBaseline


{
  "classDistribution": [
    {
      "className": "string",
      "count": 0,
      "percent": 0
    }
  ],
  "meanGeoCentroid": {
    "latitude": 0,
    "longitude": 0
  },
  "meanPredictedValue": 0,
  "meanProbabilities": [
    {
      "className": "string",
      "value": 0
    }
  ],
  "modelId": "string",
  "percentiles": [
    {
      "geoCentroid": {
        "latitude": 0,
        "longitude": 0
      },
      "percent": 0,
      "value": 0
    }
  ],
  "predictionsWarningCount": 0,
  "rowCount": 0
}

Properties¶

Name	Type	Required	Restrictions	Description
classDistribution	[ClassDistribution]	false	maxItems: 10000	Class distribution for all classes in the bucket, only for classification deployments.
meanGeoCentroid	GeoPoint	false		Geo centroid.
meanPredictedValue	number,null	false		Mean predicted value for all rows in the bucket, only for regression deployments.
meanProbabilities	[MeanProbability]	false	maxItems: 10000	Mean predicted probabilities for all classes in the bucket, only for classification deployments
modelId	string	true		ID of the model
percentiles	[Percentile]	false	maxItems: 10	Predicted value or positive class predicted probability at specific percentiles in the bucket, only for regression and binary classification deployments.
predictionsWarningCount	integer,null	false		The number of predictions with warning in the bucket
rowCount	integer,null	true		Number of rows in the bucket.

TargetDriftResponse


{
  "baselineSampleSize": 0,
  "batchId": [],
  "driftScore": 0,
  "metric": "psi",
  "modelId": "string",
  "period": {
    "end": "2019-08-24T14:15:22Z",
    "start": "2019-08-24T14:15:22Z"
  },
  "sampleSize": 0,
  "segmentAttribute": "string",
  "segmentValue": "",
  "targetName": "string",
  "type": "numeric"
}

Properties¶

Name	Type	Required	Restrictions	Description
baselineSampleSize	integer	true		sample size of the training data.
batchId	any	false		The id of the batch for which metrics are being retrieved.

oneOf

Name	Type	Required	Restrictions	Description
» anonymous	string	false		none

xor

Name	Type	Required	Restrictions	Description
» anonymous	[string]	false	maxItems: 25	none

continued

Name	Type	Required	Description
driftScore	number,null	true	drift score for the target.
metric	string	false	Metric used to calculate drift score.
modelId	string	true	id of the model for which data drift is being retrieved.
period	TimeRange	false	An object with the keys "start" and "end" defining the period.
sampleSize	integer	true	number of predictions used to compute the drift score.
segmentAttribute	string,null	false	The name of the segment on which segment analysis is being performed.
segmentValue	string,null	false	The value of the `segmentAttribute` to segment on.
targetName	string	true	name of the target feature.
type	string	true	Type of the feature.

Enumerated Values¶

Property	Value
metric	[`psi`, `kl_divergence`, `dissimilarity`, `hellinger`, `js_divergence`]
type	[`numeric`, `categorical`, `text`]

TimeRange


{
  "end": "2019-08-24T14:15:22Z",
  "start": "2019-08-24T14:15:22Z"
}

An object with the keys "start" and "end" defining the period.

Properties¶

Name	Type	Required	Restrictions	Description
end	string,null(date-time)	false		End of the period to retrieve monitoring data, defaults to the next top of the hour. Note: this field only accepts top of the hour RFC3339 datetime strings, for example: `2019-08-01T00:00:00Z`.
start	string,null(date-time)	false		Start of the period to retrieve monitoring data, defaults to 7 days ago from the end of the period. Note: this field only accepts top of the hour RFC3339 datetime strings, for example: `2019-08-01T00:00:00Z`.