Package google.cloud.aiplatform.v1beta1.schema.trainingjob.definition

AutoMlForecastingInputs

The input parameters of this TrainingJob.

AutoMlForecastingMetadata

The metadata information.

string

The name of the column that the Model is to predict values for. This column must be unavailable at forecast.

string

The name of the column that identifies the time series.

string

The name of the column that identifies time order in the time series. This column must be available at forecast.

Transformation

Each transformation will apply transform function to given input column. And the result will be used for training. When creating transformation for BigQuery Struct column, the column should be flattened using "." as the delimiter.

string

Objective function the model is optimizing towards. The training process creates a model that optimizes the value of the objective function over the validation set.

The supported optimization objectives:

• "minimize-rmse" (default) - Minimize root-mean-squared error (RMSE).

• "minimize-mae" - Minimize mean-absolute error (MAE).

• "minimize-rmsle" - Minimize root-mean-squared log error (RMSLE).

• "minimize-rmspe" - Minimize root-mean-squared percentage error (RMSPE).

• "minimize-wape-mae" - Minimize the combination of weighted absolute percentage error (WAPE) and mean-absolute-error (MAE).

• "minimize-quantile-loss" - Minimize the quantile loss at the quantiles defined in quantiles.

• "minimize-mape" - Minimize the mean absolute percentage error.

int64

Required. The train budget of creating this model, expressed in milli node hours i.e. 1,000 value in this field means 1 node hour.

The training cost of the model will not exceed this budget. The final cost will be attempted to be close to the budget, though may end up being (even) noticeably smaller - at the backend's discretion. This especially may happen when further model training ceases to provide any improvements.

If the budget is set to a value known to be insufficient to train a model for the given dataset, the training won't be attempted and will error.

The train budget must be between 1,000 and 72,000 milli node hours, inclusive.

string

Column name that should be used as the weight column. Higher values in this column give more importance to the row during model training. The column must have numeric values between 0 and 10000 inclusively; 0 means the row is ignored for training. If weight column field is not set, then all rows are assumed to have equal weight of 1.

string

Column names that should be used as attribute columns. The value of these columns does not vary as a function of time. For example, store ID or item color.

string

Names of columns that are unavailable when a forecast is requested. This column contains information for the given entity (identified by the time_series_identifier_column) that is unknown before the forecast For example, actual weather on a given day.

string

Names of columns that are available and provided when a forecast is requested. These columns contain information for the given entity (identified by the time_series_identifier_column column) that is known at forecast. For example, predicted weather for a specific day.

Granularity

Expected difference in time granularity between rows in the data.

int64

The amount of time into the future for which forecasted values for the target are returned. Expressed in number of units defined by the data_granularity field.

int64

The amount of time into the past training and prediction data is used for model training and prediction respectively. Expressed in number of units defined by the data_granularity field.

ExportEvaluatedDataItemsConfig

Configuration for exporting test set predictions to a BigQuery table. If this configuration is absent, then the export is not performed.

double

Quantiles to use for minimize-quantile-loss optimization_objective, or for probabilistic inference. Up to 5 quantiles are allowed of values between 0 and 1, exclusive. Required if the value of optimization_objective is minimize-quantile-loss. Represents the percent quantiles to use for that objective. Quantiles must be unique.

HierarchyConfig

Configuration that defines the hierarchical relationship of time series and parameters for hierarchical forecasting strategies.

WindowConfig

Config containing strategy for generating sliding windows.

string

The geographical region based on which the holiday effect is applied in modeling by adding holiday categorical array feature that include all holidays matching the date. This option only allowed when data_granularity is day. By default, holiday effect modeling is disabled. To turn it on, specify the holiday region using this option.

bool

If probabilistic inference is enabled, the model will fit a distribution that captures the uncertainty of a prediction. At inference time, the predictive distribution is used to make a point prediction that minimizes the optimization objective. For example, the mean of a predictive distribution is the point prediction that minimizes RMSE loss. If quantiles are specified, then the quantiles of the distribution are also returned. The optimization objective cannot be minimize-quantile-loss.

string

Validation options for the data validation component. The available options are:

• "fail-pipeline" - default, will validate against the validation and fail the pipeline if it fails.

• "ignore-validation" - ignore the results of the validation and continue

string

Additional experiment flags for the time series forcasting training.

string

The time granularity unit of this time period. The supported units are:

• "minute"

• "hour"

• "day"

• "week"

• "month"

• "year"

int64

The number of granularity_units between data points in the training data. If granularity_unit is minute, can be 1, 5, 10, 15, or 30. For all other values of granularity_unit, must be 1.

AutoTransformation

NumericTransformation

CategoricalTransformation

TimestampTransformation

TextTransformation

string

string

string

string

string

string

The format in which that time field is expressed. The time_format must either be one of:

• unix-seconds

• unix-milliseconds

• unix-microseconds

• unix-nanoseconds

(for respectively number of seconds, milliseconds, microseconds and nanoseconds since start of the Unix epoch);

or be written in strftime syntax.

If time_format is not set, then the default format is RFC 3339 date-time format, where time-offset = "Z" (e.g. 1985-04-12T23:20:50.52Z)

int64

Output only. The actual training cost of the model, expressed in milli node hours, i.e. 1,000 value in this field means 1 node hour. Guaranteed to not exceed the train budget.

string

BigQuery destination uri for exported evaluated examples.

AutoMlImageClassificationInputs

The input parameters of this TrainingJob.

AutoMlImageClassificationMetadata

The metadata information.

ModelType

string

The ID of the base model. If it is specified, the new model will be trained based on the base model. Otherwise, the new model will be trained from scratch. The base model must be in the same Project and Location as the new Model to train, and have the same modelType.

int64

The training budget of creating this model, expressed in milli node hours i.e. 1,000 value in this field means 1 node hour. The actual metadata.costMilliNodeHours will be equal or less than this value. If further model training ceases to provide any improvements, it will stop without using the full budget and the metadata.successfulStopReason will be model-converged. Note, node_hour = actual_hour * number_of_nodes_involved. For modelType cloud(default), the budget must be between 8,000 and 800,000 milli node hours, inclusive. The default value is 192,000 which represents one day in wall time, considering 8 nodes are used. For model types mobile-tf-low-latency-1, mobile-tf-versatile-1, mobile-tf-high-accuracy-1, the training budget must be between 1,000 and 100,000 milli node hours, inclusive. The default value is 24,000 which represents one day in wall time on a single node that is used.

bool

Use the entire training budget. This disables the early stopping feature. When false the early stopping feature is enabled, which means that AutoML Image Classification might stop training before the entire training budget has been used.

bool

If false, a single-label (multi-class) Model will be trained (i.e. assuming that for each image just up to one annotation may be applicable). If true, a multi-label Model will be trained (i.e. assuming that for each image multiple annotations may be applicable).

string

The ID of base model for upTraining. If it is specified, the new model will be upTrained based on the base model for upTraining. Otherwise, the new model will be trained from scratch. The base model for upTraining must be in the same Project and Location as the new Model to train, and have the same modelType.

int64

The actual training cost of creating this model, expressed in milli node hours, i.e. 1,000 value in this field means 1 node hour. Guaranteed to not exceed inputs.budgetMilliNodeHours.

SuccessfulStopReason

For successful job completions, this is the reason why the job has finished.

AutoMlImageObjectDetectionInputs

The input parameters of this TrainingJob.

AutoMlImageObjectDetectionMetadata

The metadata information

ModelType

int64

The training budget of creating this model, expressed in milli node hours i.e. 1,000 value in this field means 1 node hour. The actual metadata.costMilliNodeHours will be equal or less than this value. If further model training ceases to provide any improvements, it will stop without using the full budget and the metadata.successfulStopReason will be model-converged. Note, node_hour = actual_hour * number_of_nodes_involved. For modelType cloud(default), the budget must be between 20,000 and 900,000 milli node hours, inclusive. The default value is 216,000 which represents one day in wall time, considering 9 nodes are used. For model types mobile-tf-low-latency-1, mobile-tf-versatile-1, mobile-tf-high-accuracy-1 the training budget must be between 1,000 and 100,000 milli node hours, inclusive. The default value is 24,000 which represents one day in wall time on a single node that is used.

bool

Use the entire training budget. This disables the early stopping feature. When false the early stopping feature is enabled, which means that AutoML Image Object Detection might stop training before the entire training budget has been used.

string

The ID of base model for upTraining. If it is specified, the new model will be upTrained based on the base model for upTraining. Otherwise, the new model will be trained from scratch. The base model for upTraining must be in the same Project and Location as the new Model to train, and have the same modelType.

int64

The actual training cost of creating this model, expressed in milli node hours, i.e. 1,000 value in this field means 1 node hour. Guaranteed to not exceed inputs.budgetMilliNodeHours.

SuccessfulStopReason

For successful job completions, this is the reason why the job has finished.

AutoMlImageSegmentationInputs

The input parameters of this TrainingJob.

AutoMlImageSegmentationMetadata

The metadata information.

ModelType

int64

The training budget of creating this model, expressed in milli node hours i.e. 1,000 value in this field means 1 node hour. The actual metadata.costMilliNodeHours will be equal or less than this value. If further model training ceases to provide any improvements, it will stop without using the full budget and the metadata.successfulStopReason will be model-converged. Note, node_hour = actual_hour * number_of_nodes_involved. Or actual_wall_clock_hours = train_budget_milli_node_hours / (number_of_nodes_involved * 1000) For modelType cloud-high-accuracy-1(default), the budget must be between 20,000 and 2,000,000 milli node hours, inclusive. The default value is 192,000 which represents one day in wall time (1000 milli * 24 hours * 8 nodes).

string

The ID of the base model. If it is specified, the new model will be trained based on the base model. Otherwise, the new model will be trained from scratch. The base model must be in the same Project and Location as the new Model to train, and have the same modelType.

int64

The actual training cost of creating this model, expressed in milli node hours, i.e. 1,000 value in this field means 1 node hour. Guaranteed to not exceed inputs.budgetMilliNodeHours.

SuccessfulStopReason

For successful job completions, this is the reason why the job has finished.

AutoMlTablesInputs

The input parameters of this TrainingJob.

AutoMlTablesMetadata

The metadata information.

string

The type of prediction the Model is to produce. "classification" - Predict one out of multiple target values is picked for each row. "regression" - Predict a value based on its relation to other values. This type is available only to columns that contain semantically numeric values, i.e. integers or floating point number, even if stored as e.g. strings.

string

The column name of the target column that the model is to predict.

Transformation

Each transformation will apply transform function to given input column. And the result will be used for training. When creating transformation for BigQuery Struct column, the column should be flattened using "." as the delimiter.

string

Objective function the model is optimizing towards. The training process creates a model that maximizes/minimizes the value of the objective function over the validation set.

The supported optimization objectives depend on the prediction type. If the field is not set, a default objective function is used.

classification (binary): "maximize-au-roc" (default) - Maximize the area under the receiver operating characteristic (ROC) curve. "minimize-log-loss" - Minimize log loss. "maximize-au-prc" - Maximize the area under the precision-recall curve. "maximize-precision-at-recall" - Maximize precision for a specified recall value. "maximize-recall-at-precision" - Maximize recall for a specified precision value.

classification (multi-class): "minimize-log-loss" (default) - Minimize log loss.

regression: "minimize-rmse" (default) - Minimize root-mean-squared error (RMSE). "minimize-mae" - Minimize mean-absolute error (MAE). "minimize-rmsle" - Minimize root-mean-squared log error (RMSLE).

int64

Required. The train budget of creating this model, expressed in milli node hours i.e. 1,000 value in this field means 1 node hour.

The training cost of the model will not exceed this budget. The final cost will be attempted to be close to the budget, though may end up being (even) noticeably smaller - at the backend's discretion. This especially may happen when further model training ceases to provide any improvements.

If the budget is set to a value known to be insufficient to train a model for the given dataset, the training won't be attempted and will error.

The train budget must be between 1,000 and 72,000 milli node hours, inclusive.

bool

Use the entire training budget. This disables the early stopping feature. By default, the early stopping feature is enabled, which means that AutoML Tables might stop training before the entire training budget has been used.

string

Column name that should be used as the weight column. Higher values in this column give more importance to the row during model training. The column must have numeric values between 0 and 10000 inclusively; 0 means the row is ignored for training. If weight column field is not set, then all rows are assumed to have equal weight of 1.

ExportEvaluatedDataItemsConfig

Configuration for exporting test set predictions to a BigQuery table. If this configuration is absent, then the export is not performed.

string

Additional experiment flags for the Tables training pipeline.

float

Required when optimization_objective is "maximize-precision-at-recall". Must be between 0 and 1, inclusive.

float

Required when optimization_objective is "maximize-recall-at-precision". Must be between 0 and 1, inclusive.

AutoTransformation

NumericTransformation

CategoricalTransformation

TimestampTransformation

TextTransformation

NumericArrayTransformation

CategoricalArrayTransformation

TextArrayTransformation

string

string

string

string

bool

If invalid values is allowed, the training pipeline will create a boolean feature that indicated whether the value is valid. Otherwise, the training pipeline will discard the input row from trainining data.

string

bool

If invalid values is allowed, the training pipeline will create a boolean feature that indicated whether the value is valid. Otherwise, the training pipeline will discard the input row from trainining data.

string

string

string

string

The format in which that time field is expressed. The time_format must either be one of: * unix-seconds * unix-milliseconds * unix-microseconds * unix-nanoseconds (for respectively number of seconds, milliseconds, microseconds and nanoseconds since start of the Unix epoch); or be written in strftime syntax. If time_format is not set, then the default format is RFC 3339 date-time format, where time-offset = "Z" (e.g. 1985-04-12T23:20:50.52Z)

bool

If invalid values is allowed, the training pipeline will create a boolean feature that indicated whether the value is valid. Otherwise, the training pipeline will discard the input row from trainining data.

int64

Output only. The actual training cost of the model, expressed in milli node hours, i.e. 1,000 value in this field means 1 node hour. Guaranteed to not exceed the train budget.

string

BigQuery destination uri for exported evaluated examples.

AutoMlTextClassificationInputs

The input parameters of this TrainingJob.

bool

AutoMlTextExtractionInputs

The input parameters of this TrainingJob.

AutoMlTextSentimentInputs

The input parameters of this TrainingJob.

int32

A sentiment is expressed as an integer ordinal, where higher value means a more positive sentiment. The range of sentiments that will be used is between 0 and sentimentMax (inclusive on both ends), and all the values in the range must be represented in the dataset before a model can be created. Only the Annotations with this sentimentMax will be used for training. sentimentMax value must be between 1 and 10 (inclusive).

AutoMlVideoActionRecognitionInputs

The input parameters of this TrainingJob.

ModelType

AutoMlVideoClassificationInputs

The input parameters of this TrainingJob.

ModelType

AutoMlVideoObjectTrackingInputs

The input parameters of this TrainingJob.

ModelType

string

The resource name of the CustomJob that has been created to carry out this custom task.

CustomJobSpec

The input parameters of this CustomTask.

CustomJobMetadata

The metadata information.

string

URI of desired destination BigQuery table. Expected format: bq://{project_id}:{dataset_id}:{table}

If not specified, then results are exported to the following auto-created BigQuery table: {project_id}:export_evaluated_examples_{model_name}_{yyyy_MM_dd'T'HH_mm_ss_SSS'Z'}.evaluated_examples

bool

If true and an export destination is specified, then the contents of the destination are overwritten. Otherwise, if the export destination already exists, then the export operation fails.

string

A list of time series attribute column names that define the time series hierarchy. Only one level of hierarchy is supported, ex. 'region' for a hierarchy of stores or 'department' for a hierarchy of products. If multiple columns are specified, time series will be grouped by their combined values, ex. ('blue', 'large') for 'color' and 'size', up to 5 columns are accepted. If no group columns are specified, all time series are considered to be part of the same group.

double

The weight of the loss for predictions aggregated over time series in the same group.

double

The weight of the loss for predictions aggregated over the horizon for a single time series.

double

The weight of the loss for predictions aggregated over both the horizon and time series in the same hierarchy group.

string

The resource name of the HyperparameterTuningJob that has been created to carry out this HyperparameterTuning task.

string

The resource name of the CustomJob that has been created to run the best Trial of this HyperparameterTuning task.

StudySpec

Study configuration of the HyperparameterTuningJob.

CustomJobSpec

The spec of a trial job. The same spec applies to the CustomJobs created in all the trials.

int32

The desired total number of Trials.

int32

The desired number of Trials to run in parallel.

int32

The number of failed Trials that need to be seen before failing the HyperparameterTuningJob.

If set to 0, Vertex AI decides how many Trials must fail before the whole job fails.

HyperparameterTuningJobSpec

The input parameters of this HyperparameterTuningTask.

HyperparameterTuningJobMetadata

The metadata information.

Seq2SeqPlusForecastingInputs

The input parameters of this TrainingJob.

Seq2SeqPlusForecastingMetadata

The metadata information.

string

The name of the column that the Model is to predict values for. This column must be unavailable at forecast.

string

The name of the column that identifies the time series.

string

The name of the column that identifies time order in the time series. This column must be available at forecast.

Transformation

Each transformation will apply transform function to given input column. And the result will be used for training. When creating transformation for BigQuery Struct column, the column should be flattened using "." as the delimiter.

string

Objective function the model is optimizing towards. The training process creates a model that optimizes the value of the objective function over the validation set.

The supported optimization objectives:

• "minimize-rmse" (default) - Minimize root-mean-squared error (RMSE).

• "minimize-mae" - Minimize mean-absolute error (MAE).

• "minimize-rmsle" - Minimize root-mean-squared log error (RMSLE).

• "minimize-rmspe" - Minimize root-mean-squared percentage error (RMSPE).

• "minimize-wape-mae" - Minimize the combination of weighted absolute percentage error (WAPE) and mean-absolute-error (MAE).

• "minimize-quantile-loss" - Minimize the quantile loss at the quantiles defined in quantiles.

• "minimize-mape" - Minimize the mean absolute percentage error.

int64

Required. The train budget of creating this model, expressed in milli node hours i.e. 1,000 value in this field means 1 node hour.

The training cost of the model will not exceed this budget. The final cost will be attempted to be close to the budget, though may end up being (even) noticeably smaller - at the backend's discretion. This especially may happen when further model training ceases to provide any improvements.

If the budget is set to a value known to be insufficient to train a model for the given dataset, the training won't be attempted and will error.

The train budget must be between 1,000 and 72,000 milli node hours, inclusive.

string

Column name that should be used as the weight column. Higher values in this column give more importance to the row during model training. The column must have numeric values between 0 and 10000 inclusively; 0 means the row is ignored for training. If weight column field is not set, then all rows are assumed to have equal weight of 1. This column must be available at forecast.

string

Column names that should be used as attribute columns. The value of these columns does not vary as a function of time. For example, store ID or item color.

string

Names of columns that are unavailable when a forecast is requested. This column contains information for the given entity (identified by the time_series_identifier_column) that is unknown before the forecast For example, actual weather on a given day.

string

Names of columns that are available and provided when a forecast is requested. These columns contain information for the given entity (identified by the time_series_identifier_column column) that is known at forecast. For example, predicted weather for a specific day.

Granularity

Expected difference in time granularity between rows in the data.

int64

The amount of time into the future for which forecasted values for the target are returned. Expressed in number of units defined by the data_granularity field.

int64

The amount of time into the past training and prediction data is used for model training and prediction respectively. Expressed in number of units defined by the data_granularity field.

string

The geographical region based on which the holiday effect is applied in modeling by adding holiday categorical array feature that include all holidays matching the date. This option only allowed when data_granularity is day. By default, holiday effect modeling is disabled. To turn it on, specify the holiday region using this option.

ExportEvaluatedDataItemsConfig

Configuration for exporting test set predictions to a BigQuery table. If this configuration is absent, then the export is not performed.

WindowConfig

Config containing strategy for generating sliding windows.

double

Quantiles to use for minimize-quantile-loss optimization_objective. Up to 5 quantiles are allowed of values between 0 and 1, exclusive. Required if the value of optimization_objective is minimize-quantile-loss. Represents the percent quantiles to use for that objective. Quantiles must be unique.

string

Validation options for the data validation component. The available options are:

• "fail-pipeline" - default, will validate against the validation and fail the pipeline if it fails.

• "ignore-validation" - ignore the results of the validation and continue

string

Additional experiment flags for the time series forcasting training.

HierarchyConfig

Configuration that defines the hierarchical relationship of time series and parameters for hierarchical forecasting strategies.

string

The time granularity unit of this time period. The supported units are:

• "minute"

• "hour"

• "day"

• "week"

• "month"

• "year"

int64

The number of granularity_units between data points in the training data. If granularity_unit is minute, can be 1, 5, 10, 15, or 30. For all other values of granularity_unit, must be 1.

AutoTransformation

NumericTransformation

CategoricalTransformation

TimestampTransformation

TextTransformation

string

string

string

string

string

string

The format in which that time field is expressed. The time_format must either be one of:

• unix-seconds

• unix-milliseconds

• unix-microseconds

• unix-nanoseconds

(for respectively number of seconds, milliseconds, microseconds and nanoseconds since start of the Unix epoch);

or be written in strftime syntax.

If time_format is not set, then the default format is RFC 3339 date-time format, where time-offset = "Z" (e.g. 1985-04-12T23:20:50.52Z)

int64

Output only. The actual training cost of the model, expressed in milli node hours, i.e. 1,000 value in this field means 1 node hour. Guaranteed to not exceed the train budget.

string

BigQuery destination uri for exported evaluated examples.

TftForecastingInputs

The input parameters of this TrainingJob.

TftForecastingMetadata

The metadata information.

string

The name of the column that the Model is to predict values for. This column must be unavailable at forecast.

string

The name of the column that identifies the time series.

string

The name of the column that identifies time order in the time series. This column must be available at forecast.

Transformation

Each transformation will apply transform function to given input column. And the result will be used for training. When creating transformation for BigQuery Struct column, the column should be flattened using "." as the delimiter.

string

Objective function the model is optimizing towards. The training process creates a model that optimizes the value of the objective function over the validation set.

The supported optimization objectives:

• "minimize-rmse" (default) - Minimize root-mean-squared error (RMSE).

• "minimize-mae" - Minimize mean-absolute error (MAE).

• "minimize-rmsle" - Minimize root-mean-squared log error (RMSLE).

• "minimize-rmspe" - Minimize root-mean-squared percentage error (RMSPE).

• "minimize-wape-mae" - Minimize the combination of weighted absolute percentage error (WAPE) and mean-absolute-error (MAE).

• "minimize-quantile-loss" - Minimize the quantile loss at the quantiles defined in quantiles.

• "minimize-mape" - Minimize the mean absolute percentage error.

int64

Required. The train budget of creating this model, expressed in milli node hours i.e. 1,000 value in this field means 1 node hour.

The training cost of the model will not exceed this budget. The final cost will be attempted to be close to the budget, though may end up being (even) noticeably smaller - at the backend's discretion. This especially may happen when further model training ceases to provide any improvements.

If the budget is set to a value known to be insufficient to train a model for the given dataset, the training won't be attempted and will error.

The train budget must be between 1,000 and 72,000 milli node hours, inclusive.

string

Column name that should be used as the weight column. Higher values in this column give more importance to the row during model training. The column must have numeric values between 0 and 10000 inclusively; 0 means the row is ignored for training. If weight column field is not set, then all rows are assumed to have equal weight of 1. This column must be available at forecast.

string

Column names that should be used as attribute columns. The value of these columns does not vary as a function of time. For example, store ID or item color.

string

Names of columns that are unavailable when a forecast is requested. This column contains information for the given entity (identified by the time_series_identifier_column) that is unknown before the forecast For example, actual weather on a given day.

string

Names of columns that are available and provided when a forecast is requested. These columns contain information for the given entity (identified by the time_series_identifier_column column) that is known at forecast. For example, predicted weather for a specific day.

Granularity

Expected difference in time granularity between rows in the data.

int64

The amount of time into the future for which forecasted values for the target are returned. Expressed in number of units defined by the data_granularity field.

int64

The amount of time into the past training and prediction data is used for model training and prediction respectively. Expressed in number of units defined by the data_granularity field.

string

The geographical region based on which the holiday effect is applied in modeling by adding holiday categorical array feature that include all holidays matching the date. This option only allowed when data_granularity is day. By default, holiday effect modeling is disabled. To turn it on, specify the holiday region using this option.

ExportEvaluatedDataItemsConfig

Configuration for exporting test set predictions to a BigQuery table. If this configuration is absent, then the export is not performed.

WindowConfig

Config containing strategy for generating sliding windows.

double

Quantiles to use for minimize-quantile-loss optimization_objective. Up to 5 quantiles are allowed of values between 0 and 1, exclusive. Required if the value of optimization_objective is minimize-quantile-loss. Represents the percent quantiles to use for that objective. Quantiles must be unique.

string

Validation options for the data validation component. The available options are:

• "fail-pipeline" - default, will validate against the validation and fail the pipeline if it fails.

• "ignore-validation" - ignore the results of the validation and continue

string

Additional experiment flags for the time series forcasting training.

HierarchyConfig

Configuration that defines the hierarchical relationship of time series and parameters for hierarchical forecasting strategies.

string

The time granularity unit of this time period. The supported units are:

• "minute"

• "hour"

• "day"

• "week"

• "month"

• "year"

int64

The number of granularity_units between data points in the training data. If granularity_unit is minute, can be 1, 5, 10, 15, or 30. For all other values of granularity_unit, must be 1.

AutoTransformation

NumericTransformation

CategoricalTransformation

TimestampTransformation

TextTransformation

string

string

string

string

string

string

The format in which that time field is expressed. The time_format must either be one of:

• unix-seconds

• unix-milliseconds

• unix-microseconds

• unix-nanoseconds

(for respectively number of seconds, milliseconds, microseconds and nanoseconds since start of the Unix epoch);

or be written in strftime syntax.

If time_format is not set, then the default format is RFC 3339 date-time format, where time-offset = "Z" (e.g. 1985-04-12T23:20:50.52Z)

int64

Output only. The actual training cost of the model, expressed in milli node hours, i.e. 1,000 value in this field means 1 node hour. Guaranteed to not exceed the train budget.

string

BigQuery destination uri for exported evaluated examples.

string

Name of the column that should be used to generate sliding windows. The column should contain either booleans or string booleans; if the value of the row is True, generate a sliding window with the horizon starting at that row. The column will not be used as a feature in training.

int64

Stride length used to generate input examples. Within one time series, every {$STRIDE_LENGTH} rows will be used to generate a sliding window.

int64

Maximum number of windows that should be generated across all time series.