Resource: ModelMonitor
Vertex AI Model Monitoring service serves as a central hub for the analysis and visualization of data quality and performance related to models. ModelMonitor stands as a top level resource for overseeing your model monitoring tasks.
JSON representation |
---|
{ "name": string, "displayName": string, "modelMonitoringTarget": { object ( |
Fields | |
---|---|
name |
Immutable. Resource name of the ModelMonitor. Format: |
displayName |
The display name of the ModelMonitor. The name can be up to 128 characters long and can consist of any UTF-8. |
modelMonitoringTarget |
The entity that is subject to analysis. Currently only models in Vertex AI Model Registry are supported. If you want to analyze the model which is outside the Vertex AI, you could register a model in Vertex AI Model Registry using just a display name. |
trainingDataset |
Optional training dataset used to train the model. It can serve as a reference dataset to identify changes in production. |
notificationSpec |
Optional default notification spec, it can be overridden in the ModelMonitoringJob notification spec. |
outputSpec |
Optional default monitoring metrics/logs export spec, it can be overridden in the ModelMonitoringJob output spec. If not specified, a default Google Cloud Storage bucket will be created under your project. |
explanationSpec |
Optional model explanation spec. It is used for feature attribution monitoring. |
modelMonitoringSchema |
Monitoring Schema is to specify the model's features, prediction outputs and ground truth properties. It is used to extract pertinent data from the dataset and to process features based on their properties. Make sure that the schema aligns with your dataset, if it does not, we will be unable to extract data from the dataset. It is required for most models, but optional for Vertex AI AutoML Tables unless the schem information is not available. |
createTime |
Output only. timestamp when this ModelMonitor was created. A timestamp in RFC3339 UTC "Zulu" format, with nanosecond resolution and up to nine fractional digits. Examples: |
updateTime |
Output only. timestamp when this ModelMonitor was updated most recently. A timestamp in RFC3339 UTC "Zulu" format, with nanosecond resolution and up to nine fractional digits. Examples: |
satisfiesPzs |
Output only. reserved for future use. |
satisfiesPzi |
Output only. reserved for future use. |
Union field default_objective . Optional default monitoring objective, it can be overridden in the ModelMonitoringJob objective spec. default_objective can be only one of the following: |
|
tabularObjective |
Optional default tabular model monitoring objective. |
ModelMonitoringTarget
The monitoring target refers to the entity that is subject to analysis. e.g. Vertex AI Model version.
JSON representation |
---|
{ // Union field |
Fields | |
---|---|
Union field
|
|
vertexModel |
Model in Vertex AI Model Registry. |
VertexModelSource
Model in Vertex AI Model Registry.
JSON representation |
---|
{ "model": string, "modelVersionId": string } |
Fields | |
---|---|
model |
Model resource name. Format: projects/{project}/locations/{location}/models/{model}. |
modelVersionId |
Model version id. |
ModelMonitoringSchema
The Model Monitoring Schema definition.
JSON representation |
---|
{ "featureFields": [ { object ( |
Fields | |
---|---|
featureFields[] |
feature names of the model. Vertex AI will try to match the features from your dataset as follows: * For 'csv' files, the header names are required, and we will extract the corresponding feature values when the header names align with the feature names. * For 'jsonl' files, we will extract the corresponding feature values if the key names match the feature names. Note: Nested features are not supported, so please ensure your features are flattened. Ensure the feature values are scalar or an array of scalars. * For 'bigquery' dataset, we will extract the corresponding feature values if the column names match the feature names. Note: The column type can be a scalar or an array of scalars. STRUCT or JSON types are not supported. You may use SQL queries to select or aggregate the relevant features from your original table. However, ensure that the 'schema' of the query results meets our requirements. * For the Vertex AI Endpoint Request Response Logging table or Vertex AI Batch Prediction Job results. If the |
predictionFields[] |
Prediction output names of the model. The requirements are the same as the |
groundTruthFields[] |
Target /ground truth names of the model. |
FieldSchema
Schema field definition.
JSON representation |
---|
{ "name": string, "dataType": string, "repeated": boolean } |
Fields | |
---|---|
name |
Field name. |
dataType |
Supported data types are: |
repeated |
Describes if the schema field is an array of given data type. |
Methods |
|
---|---|
|
Creates a ModelMonitor. |
|
Deletes a ModelMonitor. |
|
Gets a ModelMonitor. |
|
Lists ModelMonitors in a Location. |
|
Updates a ModelMonitor. |
|
Returns the Model Monitoring alerts. |
|
Searches Model Monitoring Stats generated within a given time window. |