Getting started (console) - Amazon Personalize
Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China.

Getting started (console)

In this exercise, you use the Amazon Personalize console to create a Custom dataset group with a solution that returns movie recommendations for a given user. Before you start this exercise, review the Getting started prerequisites.

When you finish the getting started exercise, to avoid incurring unnecessary charges, follow the steps in Cleaning up resources to delete the resources you created.

In this procedure, you first create a dataset group. Next, you create an Amazon Personalize Interactions dataset in the dataset group.

To create a dataset group and a dataset

  1. Open the Amazon Personalize console at https://console.amazonaws.cn/personalize/home and sign in to your account.

  2. Choose Create dataset group.

  3. In Dataset group details, for Dataset group name, specify a name for your dataset group.

  4. For Domain choose Custom. Your screen should look similar to the following:

  5. Choose Create dataset group and continue. The Create interactions dataset page appears.

  6. On the Create interactions dataset page, for Dataset name, specify a name for your dataset.

  7. For Dataset schema, choose Create new schema. In the Schema fields section, a minimal Interactions schema is displayed. The schema matches the headers you previously added to the ratings.csv file, so you don't need to make any changes. If you haven't created the training data, see Getting started prerequisites.

  8. For Schema name, specify a name for the new schema. Your screen should look similar to the following:

  9. Choose Create dataset and continue. The Import interactions data page appears. Next, complete Step 2: Import interactions data to import interactions data.

Now that you have created a dataset, it's time to import interactions data into the dataset.

To import interactions data

  1. On the Import interactions data page, for Data import source choose Import data from S3.

  2. For Dataset import job name, specify a name for your import job.

  3. In the Additional S3 bucket policy required dialog box, if you haven't granted Amazon Personalize permissions, follow the instructions to add the required Amazon S3 bucket policy.

  4. For Data location, specify where your movie data file is stored in Amazon Simple Storage Service (S3). Use the following syntax:

    s3://<name of your S3 bucket>/<folder path>/<CSV filename>

  5. In the IAM Role section, for IAM service role, keep the default selection of Enter a custom IAM role ARN.

  6. For Custom IAM role ARN, specify the role that you created in Creating an IAM role for Amazon Personalize.

    The Dataset import job details and IAM role sections should be similar to the following:

  7. Choose Finish. The data import job starts and the Overview page is displayed. Initially, the status is Create pending (followed by Create in progress), and the Create solution button is disabled.

    The time it takes for the data to be imported depends on the size of the dataset. When the data import job has finished, the status changes to Active and the Create solution button is enabled. The Overview page should look similar to the following:

  8. After the import job has finished, choose the Create solution button. The Create solution page is displayed. Now that you have imported data, you are ready to create a solution in Step 3: Create a solution.

In this procedure, you use the dataset that you imported in Step 2: Import interactions data to train a model. A trained model is referred to as a solution version.

To create a solution

  1. On the Overview page for your dataset group, in Use custom resources choose Create solution.

  2. For Solution type, choose Item recommendation to get item recommendations for your users.

  3. For Solution name, specify a name for your solution.

  4. For Solution type choose Item recommendations.

  5. For Recipe, choose aws-user-personalization. Leave the optional Solution configuration and Advanced configuration fields unchanged.

    Your screen should look similar to the following:

  6. Choose Create and train solution. Solution version training starts and the Overview page displays.

  7. To find the training status, in the navigation pane expand Custom resources and choose Solutions and recipes.

  8. In the Solutions section, choose your solution. The details page for the solution page appears. The Solution versions page lists the status of your model.

    When the Solution version status is Active, you are ready to move to Step 4: Create a campaign.

In this procedure, you create a campaign, which deploys the solution version you created in the previous step.

To create a campaign

  1. In the navigation pane, expand Custom resources and choose Campaigns.

  2. Choose Create campaign. The Create new campaign page appears.

  3. In Campaign details, for Campaign name, specify a name for your campaign.

  4. For Solution, choose the solution you created in the previous step and for Solution version ID keep the default.

  5. For Minimum provisioned transactions per second, keep the default of 1. Leave the Campaign configuration fields unchanged.

    Your screen should look similar to the following:

  6. Choose Create campaign. Campaign creation starts and the campaign details pages with the Personalization API section displayed.

    Your screen should look similar to the following:

    Creating a campaign can take a couple minutes. After Amazon Personalize finishes creating your campaign, the page is updated to show the Test campaign results section. Your screen should look similar to the following:

In this procedure, use the campaign that you created in the previous step to get recommendations.

To get recommendations

  1. In Test campaign results, for User ID, specify a value from the ratings dataset, for example, 83. For Filter name keep the default selection of None and leave the Context fields empty.

  2. Choose Get recommendations. The Recommendations panel lists the item IDs and scores for the recommended items.

    Your screen should look similar to the following: