Changes NEW CreateExplanability API that helps you understand how attributes such as price, promotion, etc. contributes to your forecasted values; NEW CreateAutoPredictor API that trains up to 40% more accurate forecasting model, saves up to 50% of retraining time, and provides model level explainability.
Changes Predictor creation now supports selecting an accuracy metric to optimize in AutoML and hyperparameter optimization. This release adds additional accuracy metrics for predictors - AverageWeightedQuantileLoss, MAPE and MASE.
Changes Added optional field AutoMLOverrideStrategy to CreatePredictor API that allows users to customize AutoML strategy. If provided in CreatePredictor request, this field is visible in DescribePredictor and GetAccuracyMetrics responses.
Changes Updated attribute statistics in DescribeDatasetImportJob response to support Long values
Changes Added new DeleteResourceTree operation that helps in deleting all the child resources of a given resource including the given resource.
Changes This release adds EstimatedTimeRemaining minutes field to the DescribeDatasetImportJob, DescribePredictor, DescribeForecast API response which denotes the time remaining to complete the job IN_PROGRESS.
Changes Added new StopResource operation that stops Amazon Forecast resource jobs that are in progress.
Changes This release adds support for the Amazon Forecast Weather Index which can increase forecasting accuracy by automatically including weather forecasts in demand forecasts.
Changes Releasing the set of PredictorBacktestExportJob APIs which allow customers to export backtest values and item-level metrics data from Predictor training.
Changes Providing support of custom quantiles in CreatePredictor API.
Changes With this release, Amazon Forecast now supports the ability to add a tag to any resource via the launch of three new APIs: TagResouce, UntagResource and ListTagsForResource. A tag is a simple label consisting of a customer-defined key and an optional value allowing for easier resource management.
Changes This release adds two key updates to existing APIs. 1. Amazon Forecast can now generate forecasts in any quantile using the optional parameter forecastTypes in the CreateForecast API and 2. You can get additional details (metrics and relevant error messages) on your AutoML runs using the DescribePredictor and GetAccuracyMetrics APIs.
Changes Amazon Forecast is a fully managed machine learning service that makes it easy for customers to generate accurate forecasts using their historical time-series data