capabilities.work

End to End Pipeline

A 4-step recipe with Amazon Forecast. We can help you get access.

Recipe · 4 steps

End to End Pipeline

with Amazon Forecast · Artificial Intelligence Management

Chain dataset, dataset group, predictor, and forecast creation.

  1. Create the dataset that seeds the pipeline.
  2. Create a dataset group that references the new dataset by its ARN.
  3. Train a predictor using the supplied input and featurization configuration referencing the dataset group.
  4. Generate a forecast from the new predictor.
What you'll need to have ready
  • datasetName: A name for the dataset.
  • domain: The domain shared across the pipeline resources.
  • datasetType: The dataset type (e.g. TARGET_TIME_SERIES).
  • dataFrequency: The frequency of data collection.
  • schema: The schema for the dataset.
  • datasetGroupName: A name for the dataset group.
  • predictorName: A name for the predictor.
  • forecastHorizon: The number of time-steps to predict.
  • inputDataConfig: The training input configuration for the predictor.
  • featurizationConfig: The featurization configuration for the predictor.
  • forecastName: A name for the generated forecast.
For your developer

Arazzo workflow amazon-forecast:end-to-end-pipeline, 4 operations: CreateDataset → CreateDatasetGroup → CreatePredictor → CreateForecast. Specs on APIs.io.

More recipes with Amazon Forecast

Export ForecastGenerate ForecastGroup Then TrainPredict and ForecastProvision Dataset GroupProvision DatasetRegister Dataset to GroupTrain Predictor