Sign in

v1

latestOpenAPI 3.0.0Apache 2.0 License2026-07-1463278347.3 KB

Creates an Amazon Forecast predictor.

Amazon Forecast creates predictors with AutoPredictor, which involves applying the optimal combination of algorithms to each time series in your datasets. You can use CreateAutoPredictor to create new predictors or upgrade/retrain existing predictors.

Creating new predictors

The following parameters are required when creating a new predictor:

  • PredictorName - A unique name for the predictor.

  • DatasetGroupArn - The ARN of the dataset group used to train the predictor.

  • ForecastFrequency - The granularity of your forecasts (hourly, daily, weekly, etc).

  • ForecastHorizon - The number of time-steps that the model predicts. The forecast horizon is also called the prediction length.

When creating a new predictor, do not specify a value for ReferencePredictorArn.

Upgrading and retraining predictors

The following parameters are required when retraining or upgrading a predictor:

  • PredictorName - A unique name for the predictor.

  • ReferencePredictorArn - The ARN of the predictor to retrain or upgrade.

When upgrading or retraining a predictor, only specify values for the ReferencePredictorArn and PredictorName.

post/#X-Amz-Target=AmazonForecast.CreateAutoPredictor

Headers

X-Amz-Target'AmazonForecast.CreateAutoPredictor' required

Request body

PredictorNamestring required

A unique name for the predictor

ForecastHorizoninteger
<p>The number of time-steps that the model predicts. The forecast horizon is also called the prediction length.</p> <p>The maximum forecast horizon is the lesser of 500 time-steps or 1/4 of the TARGET_TIME_SERIES dataset length. If you are retraining an existing AutoPredictor, then the maximum forecast horizon is the lesser of 500 time-steps or 1/3 of the TARGET_TIME_SERIES dataset length.</p> <p>If you are upgrading to an AutoPredictor or retraining an existing AutoPredictor, you cannot update the forecast horizon parameter. You can meet this requirement by providing longer time-series in the dataset.</p>
ForecastTypesForecastType[]

The forecast types used to train a predictor. You can specify up to five forecast types. Forecast types can be quantiles from 0.01 to 0.99, by increments of 0.01 or higher. You can also specify the mean forecast with <code>mean</code>.

ForecastDimensionsName[]
<p>An array of dimension (field) names that specify how to group the generated forecast.</p> <p>For example, if you are generating forecasts for item sales across all your stores, and your dataset contains a <code>store_id</code> field, you would specify <code>store_id</code> as a dimension to group sales forecasts for each store.</p>
ForecastFrequencystring
<p>The frequency of predictions in a forecast.</p> <p>Valid intervals are an integer followed by Y (Year), M (Month), W (Week), D (Day), H (Hour), and min (Minute). For example, "1D" indicates every day and "15min" indicates every 15 minutes. You cannot specify a value that would overlap with the next larger frequency. That means, for example, you cannot specify a frequency of 60 minutes, because that is equivalent to 1 hour. The valid values for each frequency are the following:</p> <ul> <li> <p>Minute - 1-59</p> </li> <li> <p>Hour - 1-23</p> </li> <li> <p>Day - 1-6</p> </li> <li> <p>Week - 1-4</p> </li> <li> <p>Month - 1-11</p> </li> <li> <p>Year - 1</p> </li> </ul> <p>Thus, if you want every other week forecasts, specify "2W". Or, if you want quarterly forecasts, you specify "3M".</p> <p>The frequency must be greater than or equal to the TARGET_TIME_SERIES dataset frequency.</p> <p>When a RELATED_TIME_SERIES dataset is provided, the frequency must be equal to the RELATED_TIME_SERIES dataset frequency.</p>
ReferencePredictorArnstring
<p>The ARN of the predictor to retrain or upgrade. This parameter is only used when retraining or upgrading a predictor. When creating a new predictor, do not specify a value for this parameter.</p> <p>When upgrading or retraining a predictor, only specify values for the <code>ReferencePredictorArn</code> and <code>PredictorName</code>. The value for <code>PredictorName</code> must be a unique predictor name.</p>
OptimizationMetric'WAPE' | 'RMSE' | 'AverageWeightedQuantileLoss' | 'MASE' | 'MAPE'

The accuracy metric used to optimize the predictor.

ExplainPredictorboolean

Create an Explainability resource for the predictor.

Response

Success

PredictorArnstring

The Amazon Resource Name (ARN) of the predictor.

All 63 operations