Amazon Lookout for Equipment

2024/02/21 - Amazon Lookout for Equipment - 5 updated api methods

Changes  This release adds a field exposing model quality to read APIs for models. It also adds a model quality field to the API response when creating an inference scheduler.

CreateInferenceScheduler (updated) Link ¶
Changes (response)
{'ModelQuality': 'QUALITY_THRESHOLD_MET | CANNOT_DETERMINE_QUALITY | '
                 'POOR_QUALITY_DETECTED'}

Creates a scheduled inference. Scheduling an inference is setting up a continuous real-time inference plan to analyze new measurement data. When setting up the schedule, you provide an S3 bucket location for the input data, assign it a delimiter between separate entries in the data, set an offset delay if desired, and set the frequency of inferencing. You must also provide an S3 bucket location for the output data.

See also: AWS API Documentation

Request Syntax

client.create_inference_scheduler(
    ModelName='string',
    InferenceSchedulerName='string',
    DataDelayOffsetInMinutes=123,
    DataUploadFrequency='PT5M'|'PT10M'|'PT15M'|'PT30M'|'PT1H',
    DataInputConfiguration={
        'S3InputConfiguration': {
            'Bucket': 'string',
            'Prefix': 'string'
        },
        'InputTimeZoneOffset': 'string',
        'InferenceInputNameConfiguration': {
            'TimestampFormat': 'string',
            'ComponentTimestampDelimiter': 'string'
        }
    },
    DataOutputConfiguration={
        'S3OutputConfiguration': {
            'Bucket': 'string',
            'Prefix': 'string'
        },
        'KmsKeyId': 'string'
    },
    RoleArn='string',
    ServerSideKmsKeyId='string',
    ClientToken='string',
    Tags=[
        {
            'Key': 'string',
            'Value': 'string'
        },
    ]
)
type ModelName

string

param ModelName

[REQUIRED]

The name of the previously trained machine learning model being used to create the inference scheduler.

type InferenceSchedulerName

string

param InferenceSchedulerName

[REQUIRED]

The name of the inference scheduler being created.

type DataDelayOffsetInMinutes

integer

param DataDelayOffsetInMinutes

The interval (in minutes) of planned delay at the start of each inference segment. For example, if inference is set to run every ten minutes, the delay is set to five minutes and the time is 09:08. The inference scheduler will wake up at the configured interval (which, without a delay configured, would be 09:10) plus the additional five minute delay time (so 09:15) to check your Amazon S3 bucket. The delay provides a buffer for you to upload data at the same frequency, so that you don't have to stop and restart the scheduler when uploading new data.

For more information, see Understanding the inference process .

type DataUploadFrequency

string

param DataUploadFrequency

[REQUIRED]

How often data is uploaded to the source Amazon S3 bucket for the input data. The value chosen is the length of time between data uploads. For instance, if you select 5 minutes, Amazon Lookout for Equipment will upload the real-time data to the source bucket once every 5 minutes. This frequency also determines how often Amazon Lookout for Equipment runs inference on your data.

For more information, see Understanding the inference process .

type DataInputConfiguration

dict

param DataInputConfiguration

[REQUIRED]

Specifies configuration information for the input data for the inference scheduler, including delimiter, format, and dataset location.

  • S3InputConfiguration (dict) --

    Specifies configuration information for the input data for the inference, including Amazon S3 location of input data.

    • Bucket (string) -- [REQUIRED]

      The bucket containing the input dataset for the inference.

    • Prefix (string) --

      The prefix for the S3 bucket used for the input data for the inference.

  • InputTimeZoneOffset (string) --

    Indicates the difference between your time zone and Coordinated Universal Time (UTC).

  • InferenceInputNameConfiguration (dict) --

    Specifies configuration information for the input data for the inference, including timestamp format and delimiter.

    • TimestampFormat (string) --

      The format of the timestamp, whether Epoch time, or standard, with or without hyphens (-).

    • ComponentTimestampDelimiter (string) --

      Indicates the delimiter character used between items in the data.

type DataOutputConfiguration

dict

param DataOutputConfiguration

[REQUIRED]

Specifies configuration information for the output results for the inference scheduler, including the S3 location for the output.

  • S3OutputConfiguration (dict) -- [REQUIRED]

    Specifies configuration information for the output results from for the inference, output S3 location.

    • Bucket (string) -- [REQUIRED]

      The bucket containing the output results from the inference

    • Prefix (string) --

      The prefix for the S3 bucket used for the output results from the inference.

  • KmsKeyId (string) --

    The ID number for the KMS key key used to encrypt the inference output.

type RoleArn

string

param RoleArn

[REQUIRED]

The Amazon Resource Name (ARN) of a role with permission to access the data source being used for the inference.

type ServerSideKmsKeyId

string

param ServerSideKmsKeyId

Provides the identifier of the KMS key used to encrypt inference scheduler data by Amazon Lookout for Equipment.

type ClientToken

string

param ClientToken

[REQUIRED]

A unique identifier for the request. If you do not set the client request token, Amazon Lookout for Equipment generates one.

This field is autopopulated if not provided.

type Tags

list

param Tags

Any tags associated with the inference scheduler.

  • (dict) --

    A tag is a key-value pair that can be added to a resource as metadata.

    • Key (string) -- [REQUIRED]

      The key for the specified tag.

    • Value (string) -- [REQUIRED]

      The value for the specified tag.

rtype

dict

returns

Response Syntax

{
    'InferenceSchedulerArn': 'string',
    'InferenceSchedulerName': 'string',
    'Status': 'PENDING'|'RUNNING'|'STOPPING'|'STOPPED',
    'ModelQuality': 'QUALITY_THRESHOLD_MET'|'CANNOT_DETERMINE_QUALITY'|'POOR_QUALITY_DETECTED'
}

Response Structure

  • (dict) --

    • InferenceSchedulerArn (string) --

      The Amazon Resource Name (ARN) of the inference scheduler being created.

    • InferenceSchedulerName (string) --

      The name of inference scheduler being created.

    • Status (string) --

      Indicates the status of the CreateInferenceScheduler operation.

    • ModelQuality (string) --

      Provides a quality assessment for a model that uses labels. If Lookout for Equipment determines that the model quality is poor based on training metrics, the value is POOR_QUALITY_DETECTED . Otherwise, the value is QUALITY_THRESHOLD_MET .

      If the model is unlabeled, the model quality can't be assessed and the value of ModelQuality is CANNOT_DETERMINE_QUALITY . In this situation, you can get a model quality assessment by adding labels to the input dataset and retraining the model.

      For information about using labels with your models, see Understanding labeling .

      For information about improving the quality of a model, see Best practices with Amazon Lookout for Equipment .

DescribeModel (updated) Link ¶
Changes (response)
{'ModelQuality': 'QUALITY_THRESHOLD_MET | CANNOT_DETERMINE_QUALITY | '
                 'POOR_QUALITY_DETECTED'}

Provides a JSON containing the overall information about a specific machine learning model, including model name and ARN, dataset, training and evaluation information, status, and so on.

See also: AWS API Documentation

Request Syntax

client.describe_model(
    ModelName='string'
)
type ModelName

string

param ModelName

[REQUIRED]

The name of the machine learning model to be described.

rtype

dict

returns

Response Syntax

{
    'ModelName': 'string',
    'ModelArn': 'string',
    'DatasetName': 'string',
    'DatasetArn': 'string',
    'Schema': 'string',
    'LabelsInputConfiguration': {
        'S3InputConfiguration': {
            'Bucket': 'string',
            'Prefix': 'string'
        },
        'LabelGroupName': 'string'
    },
    'TrainingDataStartTime': datetime(2015, 1, 1),
    'TrainingDataEndTime': datetime(2015, 1, 1),
    'EvaluationDataStartTime': datetime(2015, 1, 1),
    'EvaluationDataEndTime': datetime(2015, 1, 1),
    'RoleArn': 'string',
    'DataPreProcessingConfiguration': {
        'TargetSamplingRate': 'PT1S'|'PT5S'|'PT10S'|'PT15S'|'PT30S'|'PT1M'|'PT5M'|'PT10M'|'PT15M'|'PT30M'|'PT1H'
    },
    'Status': 'IN_PROGRESS'|'SUCCESS'|'FAILED'|'IMPORT_IN_PROGRESS',
    'TrainingExecutionStartTime': datetime(2015, 1, 1),
    'TrainingExecutionEndTime': datetime(2015, 1, 1),
    'FailedReason': 'string',
    'ModelMetrics': 'string',
    'LastUpdatedTime': datetime(2015, 1, 1),
    'CreatedAt': datetime(2015, 1, 1),
    'ServerSideKmsKeyId': 'string',
    'OffCondition': 'string',
    'SourceModelVersionArn': 'string',
    'ImportJobStartTime': datetime(2015, 1, 1),
    'ImportJobEndTime': datetime(2015, 1, 1),
    'ActiveModelVersion': 123,
    'ActiveModelVersionArn': 'string',
    'ModelVersionActivatedAt': datetime(2015, 1, 1),
    'PreviousActiveModelVersion': 123,
    'PreviousActiveModelVersionArn': 'string',
    'PreviousModelVersionActivatedAt': datetime(2015, 1, 1),
    'PriorModelMetrics': 'string',
    'LatestScheduledRetrainingFailedReason': 'string',
    'LatestScheduledRetrainingStatus': 'IN_PROGRESS'|'SUCCESS'|'FAILED'|'IMPORT_IN_PROGRESS'|'CANCELED',
    'LatestScheduledRetrainingModelVersion': 123,
    'LatestScheduledRetrainingStartTime': datetime(2015, 1, 1),
    'LatestScheduledRetrainingAvailableDataInDays': 123,
    'NextScheduledRetrainingStartDate': datetime(2015, 1, 1),
    'AccumulatedInferenceDataStartTime': datetime(2015, 1, 1),
    'AccumulatedInferenceDataEndTime': datetime(2015, 1, 1),
    'RetrainingSchedulerStatus': 'PENDING'|'RUNNING'|'STOPPING'|'STOPPED',
    'ModelDiagnosticsOutputConfiguration': {
        'S3OutputConfiguration': {
            'Bucket': 'string',
            'Prefix': 'string'
        },
        'KmsKeyId': 'string'
    },
    'ModelQuality': 'QUALITY_THRESHOLD_MET'|'CANNOT_DETERMINE_QUALITY'|'POOR_QUALITY_DETECTED'
}

Response Structure

  • (dict) --

    • ModelName (string) --

      The name of the machine learning model being described.

    • ModelArn (string) --

      The Amazon Resource Name (ARN) of the machine learning model being described.

    • DatasetName (string) --

      The name of the dataset being used by the machine learning being described.

    • DatasetArn (string) --

      The Amazon Resouce Name (ARN) of the dataset used to create the machine learning model being described.

    • Schema (string) --

      A JSON description of the data that is in each time series dataset, including names, column names, and data types.

    • LabelsInputConfiguration (dict) --

      Specifies configuration information about the labels input, including its S3 location.

      • S3InputConfiguration (dict) --

        Contains location information for the S3 location being used for label data.

        • Bucket (string) --

          The name of the S3 bucket holding the label data.

        • Prefix (string) --

          The prefix for the S3 bucket used for the label data.

      • LabelGroupName (string) --

        The name of the label group to be used for label data.

    • TrainingDataStartTime (datetime) --

      Indicates the time reference in the dataset that was used to begin the subset of training data for the machine learning model.

    • TrainingDataEndTime (datetime) --

      Indicates the time reference in the dataset that was used to end the subset of training data for the machine learning model.

    • EvaluationDataStartTime (datetime) --

      Indicates the time reference in the dataset that was used to begin the subset of evaluation data for the machine learning model.

    • EvaluationDataEndTime (datetime) --

      Indicates the time reference in the dataset that was used to end the subset of evaluation data for the machine learning model.

    • RoleArn (string) --

      The Amazon Resource Name (ARN) of a role with permission to access the data source for the machine learning model being described.

    • DataPreProcessingConfiguration (dict) --

      The configuration is the TargetSamplingRate , which is the sampling rate of the data after post processing by Amazon Lookout for Equipment. For example, if you provide data that has been collected at a 1 second level and you want the system to resample the data at a 1 minute rate before training, the TargetSamplingRate is 1 minute.

      When providing a value for the TargetSamplingRate , you must attach the prefix "PT" to the rate you want. The value for a 1 second rate is therefore PT1S , the value for a 15 minute rate is PT15M , and the value for a 1 hour rate is PT1H

      • TargetSamplingRate (string) --

        The sampling rate of the data after post processing by Amazon Lookout for Equipment. For example, if you provide data that has been collected at a 1 second level and you want the system to resample the data at a 1 minute rate before training, the TargetSamplingRate is 1 minute.

        When providing a value for the TargetSamplingRate , you must attach the prefix "PT" to the rate you want. The value for a 1 second rate is therefore PT1S , the value for a 15 minute rate is PT15M , and the value for a 1 hour rate is PT1H

    • Status (string) --

      Specifies the current status of the model being described. Status describes the status of the most recent action of the model.

    • TrainingExecutionStartTime (datetime) --

      Indicates the time at which the training of the machine learning model began.

    • TrainingExecutionEndTime (datetime) --

      Indicates the time at which the training of the machine learning model was completed.

    • FailedReason (string) --

      If the training of the machine learning model failed, this indicates the reason for that failure.

    • ModelMetrics (string) --

      The Model Metrics show an aggregated summary of the model's performance within the evaluation time range. This is the JSON content of the metrics created when evaluating the model.

    • LastUpdatedTime (datetime) --

      Indicates the last time the machine learning model was updated. The type of update is not specified.

    • CreatedAt (datetime) --

      Indicates the time and date at which the machine learning model was created.

    • ServerSideKmsKeyId (string) --

      Provides the identifier of the KMS key used to encrypt model data by Amazon Lookout for Equipment.

    • OffCondition (string) --

      Indicates that the asset associated with this sensor has been shut off. As long as this condition is met, Lookout for Equipment will not use data from this asset for training, evaluation, or inference.

    • SourceModelVersionArn (string) --

      The Amazon Resource Name (ARN) of the source model version. This field appears if the active model version was imported.

    • ImportJobStartTime (datetime) --

      The date and time when the import job was started. This field appears if the active model version was imported.

    • ImportJobEndTime (datetime) --

      The date and time when the import job was completed. This field appears if the active model version was imported.

    • ActiveModelVersion (integer) --

      The name of the model version used by the inference schedular when running a scheduled inference execution.

    • ActiveModelVersionArn (string) --

      The Amazon Resource Name (ARN) of the model version used by the inference scheduler when running a scheduled inference execution.

    • ModelVersionActivatedAt (datetime) --

      The date the active model version was activated.

    • PreviousActiveModelVersion (integer) --

      The model version that was set as the active model version prior to the current active model version.

    • PreviousActiveModelVersionArn (string) --

      The ARN of the model version that was set as the active model version prior to the current active model version.

    • PreviousModelVersionActivatedAt (datetime) --

      The date and time when the previous active model version was activated.

    • PriorModelMetrics (string) --

      If the model version was retrained, this field shows a summary of the performance of the prior model on the new training range. You can use the information in this JSON-formatted object to compare the new model version and the prior model version.

    • LatestScheduledRetrainingFailedReason (string) --

      If the model version was generated by retraining and the training failed, this indicates the reason for that failure.

    • LatestScheduledRetrainingStatus (string) --

      Indicates the status of the most recent scheduled retraining run.

    • LatestScheduledRetrainingModelVersion (integer) --

      Indicates the most recent model version that was generated by retraining.

    • LatestScheduledRetrainingStartTime (datetime) --

      Indicates the start time of the most recent scheduled retraining run.

    • LatestScheduledRetrainingAvailableDataInDays (integer) --

      Indicates the number of days of data used in the most recent scheduled retraining run.

    • NextScheduledRetrainingStartDate (datetime) --

      Indicates the date and time that the next scheduled retraining run will start on. Lookout for Equipment truncates the time you provide to the nearest UTC day.

    • AccumulatedInferenceDataStartTime (datetime) --

      Indicates the start time of the inference data that has been accumulated.

    • AccumulatedInferenceDataEndTime (datetime) --

      Indicates the end time of the inference data that has been accumulated.

    • RetrainingSchedulerStatus (string) --

      Indicates the status of the retraining scheduler.

    • ModelDiagnosticsOutputConfiguration (dict) --

      Configuration information for the model's pointwise model diagnostics.

      • S3OutputConfiguration (dict) --

        The Amazon S3 location for the pointwise model diagnostics.

        • Bucket (string) --

          The name of the Amazon S3 bucket where the pointwise model diagnostics are located. You must be the owner of the Amazon S3 bucket.

        • Prefix (string) --

          The Amazon S3 prefix for the location of the pointwise model diagnostics. The prefix specifies the folder and evaluation result file name. (bucket ).

          When you call CreateModel or UpdateModel , specify the path within the bucket that you want Lookout for Equipment to save the model to. During training, Lookout for Equipment creates the model evaluation model as a compressed JSON file with the name model_diagnostics_results.json.gz .

          When you call DescribeModel or DescribeModelVersion , prefix contains the file path and filename of the model evaluation file.

      • KmsKeyId (string) --

        The Amazon Web Services Key Management Service (KMS) key identifier to encrypt the pointwise model diagnostics files.

    • ModelQuality (string) --

      Provides a quality assessment for a model that uses labels. If Lookout for Equipment determines that the model quality is poor based on training metrics, the value is POOR_QUALITY_DETECTED . Otherwise, the value is QUALITY_THRESHOLD_MET .

      If the model is unlabeled, the model quality can't be assessed and the value of ModelQuality is CANNOT_DETERMINE_QUALITY . In this situation, you can get a model quality assessment by adding labels to the input dataset and retraining the model.

      For information about using labels with your models, see Understanding labeling .

      For information about improving the quality of a model, see Best practices with Amazon Lookout for Equipment .

DescribeModelVersion (updated) Link ¶
Changes (response)
{'ModelQuality': 'QUALITY_THRESHOLD_MET | CANNOT_DETERMINE_QUALITY | '
                 'POOR_QUALITY_DETECTED'}

Retrieves information about a specific machine learning model version.

See also: AWS API Documentation

Request Syntax

client.describe_model_version(
    ModelName='string',
    ModelVersion=123
)
type ModelName

string

param ModelName

[REQUIRED]

The name of the machine learning model that this version belongs to.

type ModelVersion

integer

param ModelVersion

[REQUIRED]

The version of the machine learning model.

rtype

dict

returns

Response Syntax

{
    'ModelName': 'string',
    'ModelArn': 'string',
    'ModelVersion': 123,
    'ModelVersionArn': 'string',
    'Status': 'IN_PROGRESS'|'SUCCESS'|'FAILED'|'IMPORT_IN_PROGRESS'|'CANCELED',
    'SourceType': 'TRAINING'|'RETRAINING'|'IMPORT',
    'DatasetName': 'string',
    'DatasetArn': 'string',
    'Schema': 'string',
    'LabelsInputConfiguration': {
        'S3InputConfiguration': {
            'Bucket': 'string',
            'Prefix': 'string'
        },
        'LabelGroupName': 'string'
    },
    'TrainingDataStartTime': datetime(2015, 1, 1),
    'TrainingDataEndTime': datetime(2015, 1, 1),
    'EvaluationDataStartTime': datetime(2015, 1, 1),
    'EvaluationDataEndTime': datetime(2015, 1, 1),
    'RoleArn': 'string',
    'DataPreProcessingConfiguration': {
        'TargetSamplingRate': 'PT1S'|'PT5S'|'PT10S'|'PT15S'|'PT30S'|'PT1M'|'PT5M'|'PT10M'|'PT15M'|'PT30M'|'PT1H'
    },
    'TrainingExecutionStartTime': datetime(2015, 1, 1),
    'TrainingExecutionEndTime': datetime(2015, 1, 1),
    'FailedReason': 'string',
    'ModelMetrics': 'string',
    'LastUpdatedTime': datetime(2015, 1, 1),
    'CreatedAt': datetime(2015, 1, 1),
    'ServerSideKmsKeyId': 'string',
    'OffCondition': 'string',
    'SourceModelVersionArn': 'string',
    'ImportJobStartTime': datetime(2015, 1, 1),
    'ImportJobEndTime': datetime(2015, 1, 1),
    'ImportedDataSizeInBytes': 123,
    'PriorModelMetrics': 'string',
    'RetrainingAvailableDataInDays': 123,
    'AutoPromotionResult': 'MODEL_PROMOTED'|'MODEL_NOT_PROMOTED'|'RETRAINING_INTERNAL_ERROR'|'RETRAINING_CUSTOMER_ERROR'|'RETRAINING_CANCELLED',
    'AutoPromotionResultReason': 'string',
    'ModelDiagnosticsOutputConfiguration': {
        'S3OutputConfiguration': {
            'Bucket': 'string',
            'Prefix': 'string'
        },
        'KmsKeyId': 'string'
    },
    'ModelDiagnosticsResultsObject': {
        'Bucket': 'string',
        'Key': 'string'
    },
    'ModelQuality': 'QUALITY_THRESHOLD_MET'|'CANNOT_DETERMINE_QUALITY'|'POOR_QUALITY_DETECTED'
}

Response Structure

  • (dict) --

    • ModelName (string) --

      The name of the machine learning model that this version belongs to.

    • ModelArn (string) --

      The Amazon Resource Name (ARN) of the parent machine learning model that this version belong to.

    • ModelVersion (integer) --

      The version of the machine learning model.

    • ModelVersionArn (string) --

      The Amazon Resource Name (ARN) of the model version.

    • Status (string) --

      The current status of the model version.

    • SourceType (string) --

      Indicates whether this model version was created by training or by importing.

    • DatasetName (string) --

      The name of the dataset used to train the model version.

    • DatasetArn (string) --

      The Amazon Resource Name (ARN) of the dataset used to train the model version.

    • Schema (string) --

      The schema of the data used to train the model version.

    • LabelsInputConfiguration (dict) --

      Contains the configuration information for the S3 location being used to hold label data.

      • S3InputConfiguration (dict) --

        Contains location information for the S3 location being used for label data.

        • Bucket (string) --

          The name of the S3 bucket holding the label data.

        • Prefix (string) --

          The prefix for the S3 bucket used for the label data.

      • LabelGroupName (string) --

        The name of the label group to be used for label data.

    • TrainingDataStartTime (datetime) --

      The date on which the training data began being gathered. If you imported the version, this is the date that the training data in the source version began being gathered.

    • TrainingDataEndTime (datetime) --

      The date on which the training data finished being gathered. If you imported the version, this is the date that the training data in the source version finished being gathered.

    • EvaluationDataStartTime (datetime) --

      The date on which the data in the evaluation set began being gathered. If you imported the version, this is the date that the evaluation set data in the source version began being gathered.

    • EvaluationDataEndTime (datetime) --

      The date on which the data in the evaluation set began being gathered. If you imported the version, this is the date that the evaluation set data in the source version finished being gathered.

    • RoleArn (string) --

      The Amazon Resource Name (ARN) of the role that was used to train the model version.

    • DataPreProcessingConfiguration (dict) --

      The configuration is the TargetSamplingRate , which is the sampling rate of the data after post processing by Amazon Lookout for Equipment. For example, if you provide data that has been collected at a 1 second level and you want the system to resample the data at a 1 minute rate before training, the TargetSamplingRate is 1 minute.

      When providing a value for the TargetSamplingRate , you must attach the prefix "PT" to the rate you want. The value for a 1 second rate is therefore PT1S , the value for a 15 minute rate is PT15M , and the value for a 1 hour rate is PT1H

      • TargetSamplingRate (string) --

        The sampling rate of the data after post processing by Amazon Lookout for Equipment. For example, if you provide data that has been collected at a 1 second level and you want the system to resample the data at a 1 minute rate before training, the TargetSamplingRate is 1 minute.

        When providing a value for the TargetSamplingRate , you must attach the prefix "PT" to the rate you want. The value for a 1 second rate is therefore PT1S , the value for a 15 minute rate is PT15M , and the value for a 1 hour rate is PT1H

    • TrainingExecutionStartTime (datetime) --

      The time when the training of the version began.

    • TrainingExecutionEndTime (datetime) --

      The time when the training of the version completed.

    • FailedReason (string) --

      The failure message if the training of the model version failed.

    • ModelMetrics (string) --

      Shows an aggregated summary, in JSON format, of the model's performance within the evaluation time range. These metrics are created when evaluating the model.

    • LastUpdatedTime (datetime) --

      Indicates the last time the machine learning model version was updated.

    • CreatedAt (datetime) --

      Indicates the time and date at which the machine learning model version was created.

    • ServerSideKmsKeyId (string) --

      The identifier of the KMS key key used to encrypt model version data by Amazon Lookout for Equipment.

    • OffCondition (string) --

      Indicates that the asset associated with this sensor has been shut off. As long as this condition is met, Lookout for Equipment will not use data from this asset for training, evaluation, or inference.

    • SourceModelVersionArn (string) --

      If model version was imported, then this field is the arn of the source model version.

    • ImportJobStartTime (datetime) --

      The date and time when the import job began. This field appears if the model version was imported.

    • ImportJobEndTime (datetime) --

      The date and time when the import job completed. This field appears if the model version was imported.

    • ImportedDataSizeInBytes (integer) --

      The size in bytes of the imported data. This field appears if the model version was imported.

    • PriorModelMetrics (string) --

      If the model version was retrained, this field shows a summary of the performance of the prior model on the new training range. You can use the information in this JSON-formatted object to compare the new model version and the prior model version.

    • RetrainingAvailableDataInDays (integer) --

      Indicates the number of days of data used in the most recent scheduled retraining run.

    • AutoPromotionResult (string) --

      Indicates whether the model version was promoted to be the active version after retraining or if there was an error with or cancellation of the retraining.

    • AutoPromotionResultReason (string) --

      Indicates the reason for the AutoPromotionResult . For example, a model might not be promoted if its performance was worse than the active version, if there was an error during training, or if the retraining scheduler was using MANUAL promote mode. The model will be promoted in MANAGED promote mode if the performance is better than the previous model.

    • ModelDiagnosticsOutputConfiguration (dict) --

      The Amazon S3 location where Amazon Lookout for Equipment saves the pointwise model diagnostics for the model version.

      • S3OutputConfiguration (dict) --

        The Amazon S3 location for the pointwise model diagnostics.

        • Bucket (string) --

          The name of the Amazon S3 bucket where the pointwise model diagnostics are located. You must be the owner of the Amazon S3 bucket.

        • Prefix (string) --

          The Amazon S3 prefix for the location of the pointwise model diagnostics. The prefix specifies the folder and evaluation result file name. (bucket ).

          When you call CreateModel or UpdateModel , specify the path within the bucket that you want Lookout for Equipment to save the model to. During training, Lookout for Equipment creates the model evaluation model as a compressed JSON file with the name model_diagnostics_results.json.gz .

          When you call DescribeModel or DescribeModelVersion , prefix contains the file path and filename of the model evaluation file.

      • KmsKeyId (string) --

        The Amazon Web Services Key Management Service (KMS) key identifier to encrypt the pointwise model diagnostics files.

    • ModelDiagnosticsResultsObject (dict) --

      The Amazon S3 output prefix for where Lookout for Equipment saves the pointwise model diagnostics for the model version.

      • Bucket (string) --

        The name of the specific S3 bucket.

      • Key (string) --

        The Amazon Web Services Key Management Service (KMS key) key being used to encrypt the S3 object. Without this key, data in the bucket is not accessible.

    • ModelQuality (string) --

      Provides a quality assessment for a model that uses labels. If Lookout for Equipment determines that the model quality is poor based on training metrics, the value is POOR_QUALITY_DETECTED . Otherwise, the value is QUALITY_THRESHOLD_MET .

      If the model is unlabeled, the model quality can't be assessed and the value of ModelQuality is CANNOT_DETERMINE_QUALITY . In this situation, you can get a model quality assessment by adding labels to the input dataset and retraining the model.

      For information about using labels with your models, see Understanding labeling .

      For information about improving the quality of a model, see Best practices with Amazon Lookout for Equipment .

ListModelVersions (updated) Link ¶
Changes (response)
{'ModelVersionSummaries': {'ModelQuality': 'QUALITY_THRESHOLD_MET | '
                                           'CANNOT_DETERMINE_QUALITY | '
                                           'POOR_QUALITY_DETECTED'}}

Generates a list of all model versions for a given model, including the model version, model version ARN, and status. To list a subset of versions, use the MaxModelVersion and MinModelVersion fields.

See also: AWS API Documentation

Request Syntax

client.list_model_versions(
    ModelName='string',
    NextToken='string',
    MaxResults=123,
    Status='IN_PROGRESS'|'SUCCESS'|'FAILED'|'IMPORT_IN_PROGRESS'|'CANCELED',
    SourceType='TRAINING'|'RETRAINING'|'IMPORT',
    CreatedAtEndTime=datetime(2015, 1, 1),
    CreatedAtStartTime=datetime(2015, 1, 1),
    MaxModelVersion=123,
    MinModelVersion=123
)
type ModelName

string

param ModelName

[REQUIRED]

Then name of the machine learning model for which the model versions are to be listed.

type NextToken

string

param NextToken

If the total number of results exceeds the limit that the response can display, the response returns an opaque pagination token indicating where to continue the listing of machine learning model versions. Use this token in the NextToken field in the request to list the next page of results.

type MaxResults

integer

param MaxResults

Specifies the maximum number of machine learning model versions to list.

type Status

string

param Status

Filter the results based on the current status of the model version.

type SourceType

string

param SourceType

Filter the results based on the way the model version was generated.

type CreatedAtEndTime

datetime

param CreatedAtEndTime

Filter results to return all the model versions created before this time.

type CreatedAtStartTime

datetime

param CreatedAtStartTime

Filter results to return all the model versions created after this time.

type MaxModelVersion

integer

param MaxModelVersion

Specifies the highest version of the model to return in the list.

type MinModelVersion

integer

param MinModelVersion

Specifies the lowest version of the model to return in the list.

rtype

dict

returns

Response Syntax

{
    'NextToken': 'string',
    'ModelVersionSummaries': [
        {
            'ModelName': 'string',
            'ModelArn': 'string',
            'ModelVersion': 123,
            'ModelVersionArn': 'string',
            'CreatedAt': datetime(2015, 1, 1),
            'Status': 'IN_PROGRESS'|'SUCCESS'|'FAILED'|'IMPORT_IN_PROGRESS'|'CANCELED',
            'SourceType': 'TRAINING'|'RETRAINING'|'IMPORT',
            'ModelQuality': 'QUALITY_THRESHOLD_MET'|'CANNOT_DETERMINE_QUALITY'|'POOR_QUALITY_DETECTED'
        },
    ]
}

Response Structure

  • (dict) --

    • NextToken (string) --

      If the total number of results exceeds the limit that the response can display, the response returns an opaque pagination token indicating where to continue the listing of machine learning model versions. Use this token in the NextToken field in the request to list the next page of results.

    • ModelVersionSummaries (list) --

      Provides information on the specified model version, including the created time, model and dataset ARNs, and status.

      Note

      If you don't supply the ModelName request parameter, or if you supply the name of a model that doesn't exist, ListModelVersions returns an empty array in ModelVersionSummaries .

      • (dict) --

        Contains information about the specific model version.

        • ModelName (string) --

          The name of the model that this model version is a version of.

        • ModelArn (string) --

          The Amazon Resource Name (ARN) of the model that this model version is a version of.

        • ModelVersion (integer) --

          The version of the model.

        • ModelVersionArn (string) --

          The Amazon Resource Name (ARN) of the model version.

        • CreatedAt (datetime) --

          The time when this model version was created.

        • Status (string) --

          The current status of the model version.

        • SourceType (string) --

          Indicates how this model version was generated.

        • ModelQuality (string) --

          Provides a quality assessment for a model that uses labels. If Lookout for Equipment determines that the model quality is poor based on training metrics, the value is POOR_QUALITY_DETECTED . Otherwise, the value is QUALITY_THRESHOLD_MET .

          If the model is unlabeled, the model quality can't be assessed and the value of ModelQuality is CANNOT_DETERMINE_QUALITY . In this situation, you can get a model quality assessment by adding labels to the input dataset and retraining the model.

          For information about improving the quality of a model, see Best practices with Amazon Lookout for Equipment .

ListModels (updated) Link ¶
Changes (response)
{'ModelSummaries': {'ModelQuality': 'QUALITY_THRESHOLD_MET | '
                                    'CANNOT_DETERMINE_QUALITY | '
                                    'POOR_QUALITY_DETECTED'}}

Generates a list of all models in the account, including model name and ARN, dataset, and status.

See also: AWS API Documentation

Request Syntax

client.list_models(
    NextToken='string',
    MaxResults=123,
    Status='IN_PROGRESS'|'SUCCESS'|'FAILED'|'IMPORT_IN_PROGRESS',
    ModelNameBeginsWith='string',
    DatasetNameBeginsWith='string'
)
type NextToken

string

param NextToken

An opaque pagination token indicating where to continue the listing of machine learning models.

type MaxResults

integer

param MaxResults

Specifies the maximum number of machine learning models to list.

type Status

string

param Status

The status of the machine learning model.

type ModelNameBeginsWith

string

param ModelNameBeginsWith

The beginning of the name of the machine learning models being listed.

type DatasetNameBeginsWith

string

param DatasetNameBeginsWith

The beginning of the name of the dataset of the machine learning models to be listed.

rtype

dict

returns

Response Syntax

{
    'NextToken': 'string',
    'ModelSummaries': [
        {
            'ModelName': 'string',
            'ModelArn': 'string',
            'DatasetName': 'string',
            'DatasetArn': 'string',
            'Status': 'IN_PROGRESS'|'SUCCESS'|'FAILED'|'IMPORT_IN_PROGRESS',
            'CreatedAt': datetime(2015, 1, 1),
            'ActiveModelVersion': 123,
            'ActiveModelVersionArn': 'string',
            'LatestScheduledRetrainingStatus': 'IN_PROGRESS'|'SUCCESS'|'FAILED'|'IMPORT_IN_PROGRESS'|'CANCELED',
            'LatestScheduledRetrainingModelVersion': 123,
            'LatestScheduledRetrainingStartTime': datetime(2015, 1, 1),
            'NextScheduledRetrainingStartDate': datetime(2015, 1, 1),
            'RetrainingSchedulerStatus': 'PENDING'|'RUNNING'|'STOPPING'|'STOPPED',
            'ModelDiagnosticsOutputConfiguration': {
                'S3OutputConfiguration': {
                    'Bucket': 'string',
                    'Prefix': 'string'
                },
                'KmsKeyId': 'string'
            },
            'ModelQuality': 'QUALITY_THRESHOLD_MET'|'CANNOT_DETERMINE_QUALITY'|'POOR_QUALITY_DETECTED'
        },
    ]
}

Response Structure

  • (dict) --

    • NextToken (string) --

      An opaque pagination token indicating where to continue the listing of machine learning models.

    • ModelSummaries (list) --

      Provides information on the specified model, including created time, model and dataset ARNs, and status.

      • (dict) --

        Provides information about the specified machine learning model, including dataset and model names and ARNs, as well as status.

        • ModelName (string) --

          The name of the machine learning model.

        • ModelArn (string) --

          The Amazon Resource Name (ARN) of the machine learning model.

        • DatasetName (string) --

          The name of the dataset being used for the machine learning model.

        • DatasetArn (string) --

          The Amazon Resource Name (ARN) of the dataset used to create the model.

        • Status (string) --

          Indicates the status of the machine learning model.

        • CreatedAt (datetime) --

          The time at which the specific model was created.

        • ActiveModelVersion (integer) --

          The model version that the inference scheduler uses to run an inference execution.

        • ActiveModelVersionArn (string) --

          The Amazon Resource Name (ARN) of the model version that is set as active. The active model version is the model version that the inference scheduler uses to run an inference execution.

        • LatestScheduledRetrainingStatus (string) --

          Indicates the status of the most recent scheduled retraining run.

        • LatestScheduledRetrainingModelVersion (integer) --

          Indicates the most recent model version that was generated by retraining.

        • LatestScheduledRetrainingStartTime (datetime) --

          Indicates the start time of the most recent scheduled retraining run.

        • NextScheduledRetrainingStartDate (datetime) --

          Indicates the date that the next scheduled retraining run will start on. Lookout for Equipment truncates the time you provide to the nearest UTC day .

        • RetrainingSchedulerStatus (string) --

          Indicates the status of the retraining scheduler.

        • ModelDiagnosticsOutputConfiguration (dict) --

          Output configuration information for the pointwise model diagnostics for an Amazon Lookout for Equipment model.

          • S3OutputConfiguration (dict) --

            The Amazon S3 location for the pointwise model diagnostics.

            • Bucket (string) --

              The name of the Amazon S3 bucket where the pointwise model diagnostics are located. You must be the owner of the Amazon S3 bucket.

            • Prefix (string) --

              The Amazon S3 prefix for the location of the pointwise model diagnostics. The prefix specifies the folder and evaluation result file name. (bucket ).

              When you call CreateModel or UpdateModel , specify the path within the bucket that you want Lookout for Equipment to save the model to. During training, Lookout for Equipment creates the model evaluation model as a compressed JSON file with the name model_diagnostics_results.json.gz .

              When you call DescribeModel or DescribeModelVersion , prefix contains the file path and filename of the model evaluation file.

          • KmsKeyId (string) --

            The Amazon Web Services Key Management Service (KMS) key identifier to encrypt the pointwise model diagnostics files.

        • ModelQuality (string) --

          Provides a quality assessment for a model that uses labels. If Lookout for Equipment determines that the model quality is poor based on training metrics, the value is POOR_QUALITY_DETECTED . Otherwise, the value is QUALITY_THRESHOLD_MET .

          If the model is unlabeled, the model quality can't be assessed and the value of ModelQuality is CANNOT_DETERMINE_QUALITY . In this situation, you can get a model quality assessment by adding labels to the input dataset and retraining the model.

          For information about using labels with your models, see Understanding labeling .

          For information about improving the quality of a model, see Best practices with Amazon Lookout for Equipment .