Custom resources for training and deploying Amazon Personalize models

Create and configure a solution: Customize solution parameters and recipe-specific hyperparameters so the model meets your specific business needs. By default, new solution versions use automatic training to create solution versions at a configurable frequency. The default frequency is every 7 days. For more information about configuring a solution, see Configuring a solution. For more information about custom recipes in Amazon Personalize, see Choosing a recipe.

Create a solution version (for solutions that don't use automatic training): For solutions that use automatic training, solution version creation starts automatically after your solution is active. For solutions that use manual training, you manually create a solution version. The solution version generates Amazon Personalize recommendations or user segments. For more information about manually creating a solution version, see Manually creating a solution version. To stop solution version creation, navigate to the solution version details page and choose Stop. For more information, see Stopping the creation of a solution version.

Evaluate the solution version – Use the metrics Amazon Personalize generates from the new solution version to evaluate the performance of the model. See Evaluating an Amazon Personalize solution version with metrics.

Deploy the solution version with a campaign (only for real-time recommendations): Create a campaign to deploy your solution version. You use the campaign when you request real-time recommendations. If you are getting batch recommendations, you don't need to create a campaign. For more information, see Deploying an Amazon Personalize solution version with a campaign. If you want to change an existing campaign's settings, such as enabling metadata in recommendations, you must update your campaign. For more information, see Updating an Amazon Personalize campaign's configuration.