Guidelines and Quotas - Amazon Forecast
Supported AWS RegionsComplianceService QuotasConditions and Restrictions

Amazon Forecast is no longer available to new customers. Existing customers of Amazon Forecast can continue to use the service as normal. Learn more"

Guidelines and Quotas

The following sections contain information about Amazon Forecast guidelines and quotas.

Supported AWS Regions

For a list of AWS Regions that support Forecast, see AWS Regions and Endpoints in the Amazon Web Services General Reference.

Compliance

For more information about Forecast compliance programs, see AWS Compliance, AWS Compliance Programs, and AWS Services in Scope by Compliance Program.

Service Quotas

Note

To request an increase for adjustable quotas, use the Service Quotas console and follow the steps in the Requesting a quota increase section of the Service Quotas User Guide.

Forecast has the following service quotas.

Quotas Imposed by the CreateDatasetImportJob API
Resource Default Quota Adjustable
Maximum number of files in your Amazon S3 bucket 10,000 No
Maximum cumulative size of all files in your Amazon S3 bucket 30 GB Yes
Maximum number of datasets in a dataset group 3 (1 for each type) No
Maximum number of rows in a dataset

3 billion

Note: the quota for the ap-south-1 region is 1 billion.

 Yes

Maximum number of columns in a target time series dataset

(required columns + additional forecast dimensions)

 13 (3 + 10) No

Maximum number of columns in a related time series dataset

(required columns + additional forecast dimensions + related features)

 25 (2 + 10 + 13) No
Maximum number of columns in an item metadata dataset 10 No
Maximum number of columns in any other dataset 36 No
Quotas Imposed by the CreatePredictor API
Resource Default Quota Adjustable
Maximum number of backtest windows (EvaluationParameters) 5 No

Maximum number of time series per predictor

(number of items X number of unique values across forecast dimensions in the target time series dataset)

5,000,000 across all target time series items and dimensions.

Note: the quota for the ap-south-1 region is 1,000,000 .

If you exceed 100,000 items, Forecast supports yearly, monthly, weekly, and daily frequencies instead of more granular frequencies (such as hourly).

 Yes
Maximum forecast horizon

CNN-QR, DeepAR+, AutoML: The lesser of 500 data points or 1/3 of the target time series dataset length

ETS, NPTS, Prophet, ARIMA: The lesser of 500 data points or the length of the target time series dataset minus one.

 No
General Resource Quotas
Resource Default Quota Adjustable
Maximum parallel running CreateDatasetImportJob tasks 3 Yes
Maximum parallel running CreatePredictor tasks 3 Yes
Maximum parallel running CreatePredictor tasks using AutoML 3 Yes
Maximum parallel running CreateAutoPredictor tasks 3 No
Maximum parallel running CreateExplainability tasks 3 No
Maximum parallel running CreateExplainabilityExport tasks 3 No
Maximum parallel running CreatePredictorBacktestExportJob tasks 3 Yes
Maximum parallel running CreateForecast tasks 3 Yes
Maximum parallel running CreateForecastExportJob tasks 3 Yes
Maximum parallel running StopResource tasks per resource type 3 Yes
Maximum number of datasets 1500 Yes
Maximum number of dataset groups 500 Yes
Maximum number of dataset import jobs 1000 Yes
Maximum number of predictors 500 Yes
Maximum number of AutoPredictors 500 No
Maximum number of predictor backtest export jobs 1000 Yes
Maximum number of forecasts

100

 Yes
Maximum number of forecast export jobs 1000 Yes
Maximum time for which a forecast can be queried on console or QueryForecast API 30 days No
Maximum number of tags you can add to a resource 50 No
Maximum parallel running QueryForecast API tasks

10 forecasts, including 5 created with large datasets (anything over 20GB or 100,000 items).

If you have more than 5 forecasts created with large datasets, QueryForecast can access only the 5 most recent large dataset forecasts.

 No
Maximum number of Explainabilities 1000 No
Maximum number of Explainability Export jobs 1000 No
What-if Analysis Quotas
Resource Default Quota Adjustable
Maximum parallel running CreateWhatIfAnalysis tasks 3 Yes
Maximum number of what-if analyses 500 Yes
Maximum parallel running CreateWhatIfForecast tasks 3 Yes
Maximum number of what-if forecasts 100 Yes
Maximum parallel running CreateWhatIfForecastExport tasks 3 Yes
Maximum number of what-if forecast exports 1000 Yes
Maximum number of what-if forecasts in an export job 3 No

Conditions and Restrictions

The following conditions and restrictions apply when using the Weather Index:

  • Available algorithms: If using a legacy predictor, the Weather Index can be enabled when you train a predictor with the CNN-QR, DeepAR+, and Prophet algorithms. The Weather Index is not applied to ARIMA, ETS, and NPTS.

  • Forecast frequency: The valid forecast frequencies are Minutely, Hourly, and Daily.

  • Forecast horizon: The forecast horizon cannot span further than 14 days into the future. For forecast horizon limits for each forecast frequency, refer to the list below:

    • 1 minute - 500

    • 5 minutes - 500

    • 10 minutes - 500

    • 15 minutes - 500

    • Hourly - 330

    • Daily - 14

  • Time series length: When training a model with the Weather Index, Forecast truncates all time series datasets with timestamps before the start date of the Forecast weather dataset featurization. The Forecast weather dataset featurization contains the following start dates:

    • US region: July 2, 2018

    • Europe region: July 2, 2018

    • Asia Pacific region: July 2, 2018

    • Canada region: July 2, 2019

    • South America region: January 2, 2020

    • Central America region: September 2, 2020

    • Africa & Middle East region: March 25, 2021

    With the Weather Index enabled, data points with timestamps before the start date will not be used during predictor training.

  • Number of locations: The target time series dataset cannot exceed 2000 unique locations.

  • Region bounds: All items in your datasets must be located within a single region.

  • Minimum time series length: Due to additional data requirements when testing the Weather Index, the minimum length for a time series dataset is:

    3 × ForecastHorizon + (BacktestWindows + 1) × BacktestWindowOffset

    If your time series datasets do not meet this requirement, consider decreasing the following:

    • ForecastHorizon - Shorten your forecast horizon.

    • BacktestWindowOffset - Shorten the length of the testing set during backtesting.

    • BacktestWindows - Reduce the number of backtests.