Step 2: Creating a dataset and a schema

After you have completed Step 1: Creating a Custom dataset group, you are ready to create a dataset. Datasets are Amazon Personalize containers for data. When you create a dataset, you also create a schema for the dataset. A schema tells Amazon Personalize about the structure of your data and allows Amazon Personalize to parse the data.

You create datasets with the Amazon Personalize console, AWS Command Line Interface (AWS CLI), or AWS SDKs. For information about the different types of datasets, and dataset and schema requirements, see Datasets and schemas.

Creating a dataset and a schema (console)

If this is your first dataset in your dataset group, your first dataset type will be an Interactions dataset. To create your Interactions dataset in the console, specify the dataset name and then specify a JSON schema in Avro format. If it is not your first dataset in this dataset group, choose the dataset type and then specify a name and a schema.

For information on Amazon Personalize datasets and schema requirements, see Datasets and schemas.

Note

If you just completed Step 1: Creating a Custom dataset group and you are already on the user-item interaction page, skip to step 4 in this procedure.

To create a dataset and a schema

1. Open the Amazon Personalize console at https://console.aws.amazon.com/personalize/home and sign in to your account.

2. On the Dataset groups page, choose the dataset group you created in Step 1: Creating a Custom dataset group. This displays the dataset group Dashboard.

3. In the Upload datasets section, for the type of dataset that you want to import (Amazon Personalize datasets include Interactions, Users, or Items), choose Import. The Configure < dataset type > page is displayed.

4. In Dataset details, for Dataset name, specify a name for your dataset.

5. In Schema details, for Schema selection, either choose an existing schema or choose Create new schema.

6. If you are creating a new schema, for Schema definition, paste in the schema JSON that matches your data. Use the examples found in Datasets and schemas as a guide. After you create a schema, you can't make changes to the schema.

7. For New schema name, specify a name for the new schema.

8. For Tags, optionally add any tags. For more information about tagging Amazon Personalize resources, see Tagging Amazon Personalize resources.

9. Choose Next and follow the instructions in Step 3: Importing your historical data to import your data.

Creating a dataset and a schema (AWS CLI)

To create a dataset and a schema using the AWS CLI, you first define a schema in Avro format and add it to Amazon Personalize using the CreateSchema operation. Then create a dataset using the CreateDataset operation. For information on Amazon Personalize datasets and schema requirements, see Datasets and schemas.

To create a schema and dataset

1. Create a schema file in Avro format and save it as a JSON file. This file should be based on the type of dataset, such as Interactions, you are creating.

   The schema must match the columns in your data and the schema name must match one of the three types of datasets recognized by Amazon Personalize. The following is an example of a minimal Interactions dataset schema. For more examples, see Datasets and schemas.

   {
     "type": "record",
     "name": "Interactions",
     "namespace": "com.amazonaws.personalize.schema",
     "fields": [
         {
             "name": "USER_ID",
             "type": "string"
         },
         {
             "name": "ITEM_ID",
             "type": "string"
         },
         {
             "name": "TIMESTAMP",
             "type": "long"
         }
     ],
     "version": "1.0"
   }

2. Create a schema in Amazon Personalize by running the following command. After you create a schema, you can't make changes to the schema. Replace schemaName with the name of the schema, and replace file://SchemaName.json with the location of the JSON file you created in the previous step. The example shows the file as belonging to the current folder. For more information about the API, see CreateSchema.

   aws personalize create-schema \
     --name SchemaName \
     --schema file://SchemaName.json

   The schema Amazon Resource Name (ARN) is displayed, as shown in the following example:

   {
     "schemaArn": "arn:aws:personalize:us-west-2:acct-id:schema/SchemaName"
   }

3. Create an empty dataset by running the following command. Provide the dataset group Amazon Resource Name (ARN) from Creating a dataset group (AWS CLI) and schema ARN from the previous step. The dataset-type must match the schema name from the previous step. For more information about the API, see CreateDataset.

   aws personalize create-dataset \
     --name Dataset Name \
     --dataset-group-arn Dataset Group ARN \
     --dataset-type Dataset Type \
     --schema-arn Schema Arn
   

   The dataset ARN is displayed, as shown in the following example.

   {
     "datasetArn": "arn:aws:personalize:us-west-2:acct-id:dataset/DatasetName/INTERACTIONS"
   }

4. Record the dataset ARN for later use. After you have created a dataset, you are ready to import your training data. See Step 3: Importing your historical data.

Creating a dataset and a schema (AWS SDKs)

To create a dataset and a schema using the AWS SDKs, you first define a schema in Avro format and add it to Amazon Personalize using the CreateSchema operation. After you create a schema, you can't make changes to the schema. Then create a dataset using the CreateDataset operation. For information on Amazon Personalize datasets and schema requirements, see Datasets and schemas.

To create a schema and a dataset

1. Create a schema file in Avro format and save it as a JSON file in your working directory.

   The schema must match the columns in your data and the schema name must match one of the three types of datasets recognized by Amazon Personalize. The following is an example of a minimal Interactions dataset schema. For more examples, see Datasets and schemas.

   {
     "type": "record",
     "name": "Interactions",
     "namespace": "com.amazonaws.personalize.schema",
     "fields": [
         {
             "name": "USER_ID",
             "type": "string"
         },
         {
             "name": "ITEM_ID",
             "type": "string"
         },
         {
             "name": "TIMESTAMP",
             "type": "long"
         }
     ],
     "version": "1.0"
   }

2. Create the schema using the CreateSchema API operation. The following code shows how to create a schema. Specify the name for your schema and the file path for your schema JSON file.

   SDK for Python (Boto3)

   import boto3
   
   personalize = boto3.client('personalize')
   
   with open('schemaFile.json') as f:
       createSchemaResponse = personalize.create_schema(
           name = 'schema name',
           schema = f.read()
       )
   
   schema_arn = createSchemaResponse['schemaArn']
   
   print('Schema ARN:' + schema_arn )

   SDK for Java 2.x

   public static String createSchema(PersonalizeClient personalizeClient, String schemaName, String filePath) {
       
       String schema = null;
       
       try {
           schema = new String(Files.readAllBytes(Paths.get(filePath)));
       } catch (IOException e) {
           System.out.println(e.getMessage());
       }
       
       try {
           CreateSchemaRequest createSchemaRequest = CreateSchemaRequest.builder()
                   .name(schemaName)
                   .schema(schema)
                   .build();
   
           String schemaArn = personalizeClient.createSchema(createSchemaRequest).schemaArn();    
           System.out.println("Schema arn: " + schemaArn);
   
           return schemaArn;
   
       } catch(PersonalizeException e) {
           System.err.println(e.awsErrorDetails().errorMessage());
           System.exit(1);
       }
       return "";
   }

   SDK for JavaScript v3

   // Get service clients module and commands using ES6 syntax.
   import { CreateSchemaCommand } from
     "@aws-sdk/client-personalize";
   import { personalizeClient } from "./libs/personalizeClients.js";
   
   // Or, create the client here.
   // const personalizeClient = new PersonalizeClient({ region: "REGION"});
   
   import fs from 'fs';
   
   let schemaFilePath = "SCHEMA_PATH";
   let mySchema = "";
   
   try {
     mySchema = fs.readFileSync(schemaFilePath).toString();
   } catch (err) {
     mySchema = 'TEST' // For unit tests.
   }
   // Set the schema parameters.
   export const createSchemaParam = {
     name: 'NAME', /* required */
     schema: mySchema /* required */
   };
   
   export const run = async () => {
     try {
       const response = await personalizeClient.send(new CreateSchemaCommand(createSchemaParam));
       console.log("Success", response);
       return response; // For unit tests.
     } catch (err) {
       console.log("Error", err);
     }
   };
   run();
   

   Amazon Personalize returns the ARN of the new schema. Record it because you'll need it in the next step.

3. Create a dataset using the CreateDataset operation. The following code shows how to create a dataset. Specify the Amazon Resource Name (ARN) of your dataset group, the schema ARN from the previous step, and specify the dataset type (Interactions, Users, or Items). For information about the different types of datasets, see Datasets and schemas.

   SDK for Python (Boto3)

   import boto3
   
   personalize = boto3.client('personalize')
   
   response = personalize.create_dataset(
       name = 'datase_name',
       schemaArn = 'schema_arn',
       datasetGroupArn = 'dataset_group_arn',
       datasetType = 'dataset_type'
   )
   
   print ('Dataset Arn: ' + response['datasetArn'])

   SDK for Java 2.x

   public static String createDataset(PersonalizeClient personalizeClient, 
                                       String datasetName, 
                                       String datasetGroupArn, 
                                       String datasetType, 
                                       String schemaArn) {
       try {
           CreateDatasetRequest request = CreateDatasetRequest.builder()
                   .name(datasetName)
                   .datasetGroupArn(datasetGroupArn)
                   .datasetType(datasetType)
                   .schemaArn(schemaArn).build();
       
           String datasetArn = personalizeClient.createDataset(request).datasetArn();
           System.out.println("Dataset " + datasetName + " created. Dataset ARN: " + datasetArn);
           
           return datasetArn;
           
       } catch(PersonalizeException e) {
           System.err.println(e.awsErrorDetails().errorMessage());
           System.exit(1);
       }
       return "";
   }

   SDK for JavaScript v3

   // Get service clients module and commands using ES6 syntax.
   import { CreateDatasetCommand } from
     "@aws-sdk/client-personalize";
   import { personalizeClient } from "./libs/personalizeClients.js";
   
   // Or, create the client here.
   // const personalizeClient = new PersonalizeClient({ region: "REGION"});
   
   // Set the dataset's parameters.
   export const createDatasetParam = {
     datasetGroupArn: 'DATASET_GROUP_ARN', /* required */
     datasetType: 'DATASET_TYPE', /* required */
     name: 'NAME', /* required */
     schemaArn: 'SCHEMA_ARN' /* required */
   }
   
   export const run = async () => {
     try {
       const response = await personalizeClient.send(new CreateDatasetCommand(createDatasetParam));
       console.log("Success", response);
       return response; // For unit tests.
     } catch (err) {
       console.log("Error", err);
     }
   };
   run();
   

   After you have created a dataset, you are ready to import your training data. See Step 3: Importing your historical data.