Input Data steps are coloured pink. They are used to import input data for model training and map datasets to a specific application blocks, for example Amazon Personalize.

The step also lets you specify the instance size that will be required to perform the processing.


Contents


Selecting an Input Data Step

When editing or adding a workflow, click on the 'Plus' icon and choose Input Data Step.



Input Step Configuration


Step Name

The name must be unique within the workflow you are creating.

Only alphanumeric characters, hyphens, underscores and dots (.) are allowed.

The name must start and end with a letter.


Application Type

Choose the required application from the dropdown menu:




Map Input Dataset

Based on the application type that you have chosen, the system will show the standard input datasets that need to be mapped.

Once you have selected a dataset, a new window appears that will enable you to select a schema and Redshift table where your input data is saved.

Amazon Forecast Map Input Dataset Options

For a guide to mapping the input dataset for Amazon Forecast, see <link>


Amazon Personalize Map Input Dataset Options

For a guide to mapping the input dataset for Amazon Personalize, see Input and map data from your Redshift table.



Data Collection Frequency (Amazon Forecast only)

This is the frequency at which data will be collected from the Redshift table.

Choose from:

  • Yearly
  • Monthly
  • Weekly
  • Daily
  • Hourly
  • 30 minutes


Instance Size

Choose from:

General Purpose

Memory Intensive

CPU Intensive

  • Pico
    (0.125CPU, 0.125GB Memory)

  • Nano
    (0.25CPU, 0.5GB Memory)

  • Micro
    (0.5CPU - 1 GB Memory)

  • Small
    (1CPU - 4 GB Memory)

  • Regular
    (2CPU, 4GB RAM)

  • Medium
    (2CPUs, 8GB RAM)

  • Large
    (4CPUs, 16GB RAM)

  • XLarge
    (8CPUs, 32GB RAM)

  • 2XLarge
    (16CPUs, 64GB RAM)

  • 4XLarge
    (32CPUs, 128GB RAM)

  • Medium
    (4CPUs, 8GB RAM)

  • Large
    (8CPUs, 16GB RAM)

Dependency

This lists all of the other steps within the workflow.

Select any steps that must be successfully run before this step can run.