ModelMonitoringInput

Model monitoring data input spec.

Fields
Union field dataset. Dataset source. dataset can be only one of the following:
columnizedDataset object (ModelMonitoringDataset)

Columnized dataset.

batchPredictionOutput object (BatchPredictionOutput)

Vertex AI Batch prediction Job.

vertexEndpointLogs object (VertexEndpointLogs)

Vertex AI Endpoint request & response logging.

Union field time_spec. Time specification for the dataset. time_spec can be only one of the following:
timeInterval object (Interval)

The time interval (pair of startTime and endTime) for which results should be returned.

timeOffset object (TimeOffset)

The time offset setting for which results should be returned.

JSON representation
{

  // Union field dataset can be only one of the following:
  "columnizedDataset": {
    object (ModelMonitoringDataset)
  },
  "batchPredictionOutput": {
    object (BatchPredictionOutput)
  },
  "vertexEndpointLogs": {
    object (VertexEndpointLogs)
  }
  // End of list of possible types for union field dataset.

  // Union field time_spec can be only one of the following:
  "timeInterval": {
    object (Interval)
  },
  "timeOffset": {
    object (TimeOffset)
  }
  // End of list of possible types for union field time_spec.
}

ModelMonitoringDataset

Input dataset spec.

Fields
timestampField string

The timestamp field. Usually for serving data.

Union field data_location. Choose one of supported data location for columnized dataset. data_location can be only one of the following:
vertexDataset string

Resource name of the Vertex AI managed dataset.

gcsSource object (ModelMonitoringGcsSource)

Google Cloud Storage data source.

bigquerySource object (ModelMonitoringBigQuerySource)

BigQuery data source.

JSON representation
{
  "timestampField": string,

  // Union field data_location can be only one of the following:
  "vertexDataset": string,
  "gcsSource": {
    object (ModelMonitoringGcsSource)
  },
  "bigquerySource": {
    object (ModelMonitoringBigQuerySource)
  }
  // End of list of possible types for union field data_location.
}

ModelMonitoringGcsSource

Dataset spec for data stored in Google Cloud Storage.

Fields
gcsUri string

Google Cloud Storage URI to the input file(s). May contain wildcards. For more information on wildcards, see https://cloud.google.com/storage/docs/gsutil/addlhelp/WildcardNames.

format enum (DataFormat)

data format of the dataset.

JSON representation
{
  "gcsUri": string,
  "format": enum (DataFormat)
}

DataFormat

Supported data format.

Enums
DATA_FORMAT_UNSPECIFIED data format unspecified, used when this field is unset.
CSV CSV files.
TF_RECORD TfRecord files
JSONL JsonL files.

ModelMonitoringBigQuerySource

Dataset spec for data sotred in BigQuery.

Fields

Union field connection.

connection can be only one of the following:

tableUri string

BigQuery URI to a table, up to 2000 characters long. All the columns in the table will be selected. Accepted forms:

  • BigQuery path. For example: bq://projectId.bqDatasetId.bqTableId.
query string

Standard SQL to be used instead of the tableUri.

JSON representation
{

  // Union field connection can be only one of the following:
  "tableUri": string,
  "query": string
  // End of list of possible types for union field connection.
}

BatchPredictionOutput

data from Vertex AI Batch prediction job output.

Fields
batchPredictionJob string

Vertex AI Batch prediction job resource name. The job must match the model version specified in [ModelMonitor].[modelMonitoringTarget].

JSON representation
{
  "batchPredictionJob": string
}

VertexEndpointLogs

data from Vertex AI Endpoint request response logging.

Fields
endpoints[] string

List of endpoint resource names. The endpoints must enable the logging with the [Endpoint].[requestResponseLoggingConfig], and must contain the deployed model corresponding to the model version specified in [ModelMonitor].[modelMonitoringTarget].

JSON representation
{
  "endpoints": [
    string
  ]
}

TimeOffset

time offset setting.

Fields
offset string

[offset] is the time difference from the cut-off time. For scheduled jobs, the cut-off time is the scheduled time. For non-scheduled jobs, it's the time when the job was created. Currently we support the following format: 'w|W': Week, 'd|D': Day, 'h|H': Hour E.g. '1h' stands for 1 hour, '2d' stands for 2 days.

window string

[window] refers to the scope of data selected for analysis. It allows you to specify the quantity of data you wish to examine. Currently we support the following format: 'w|W': Week, 'd|D': Day, 'h|H': Hour E.g. '1h' stands for 1 hour, '2d' stands for 2 days.

JSON representation
{
  "offset": string,
  "window": string
}