FeatureGroup ​
The FeatureGroup resource lets you manage AWS SageMaker FeatureGroups for organizing and storing features for machine learning models.
Minimal Example ​
Create a basic FeatureGroup with required properties and some common optional configurations.
ts
import AWS from "alchemy/aws/control";
const featureGroup = await AWS.SageMaker.FeatureGroup("myFeatureGroup", {
FeatureGroupName: "CustomerFeatures",
RecordIdentifierFeatureName: "customerId",
EventTimeFeatureName: "eventTime",
FeatureDefinitions: [
{
FeatureName: "customerId",
FeatureType: "String"
},
{
FeatureName: "eventTime",
FeatureType: "Timestamp"
},
{
FeatureName: "purchaseAmount",
FeatureType: "Number"
}
],
Description: "Feature group for storing customer purchase information.",
OnlineStoreConfig: {
EnableOnlineStore: true
},
OfflineStoreConfig: {
S3StorageConfig: {
S3Uri: "s3://my-bucket/offline-store",
KmsKeyId: "arn:aws:kms:us-east-1:123456789012:key/abcd1234-56ef-78gh-90ij-klmnopqrst"
}
}
});
Advanced Configuration ​
Configure a FeatureGroup with a custom throughput configuration and IAM role.
ts
const advancedFeatureGroup = await AWS.SageMaker.FeatureGroup("advancedFeatureGroup", {
FeatureGroupName: "AdvancedCustomerFeatures",
RecordIdentifierFeatureName: "customerId",
EventTimeFeatureName: "eventTime",
FeatureDefinitions: [
{
FeatureName: "customerId",
FeatureType: "String"
},
{
FeatureName: "eventTime",
FeatureType: "Timestamp"
},
{
FeatureName: "loyaltyPoints",
FeatureType: "Number"
}
],
Description: "Feature group for advanced customer features.",
RoleArn: "arn:aws:iam::123456789012:role/SageMakerExecutionRole",
ThroughputConfig: {
VolumeSizeInGB: 5,
KmsKeyId: "arn:aws:kms:us-east-1:123456789012:key/abcd1234-56ef-78gh-90ij-klmnopqrst"
}
});
Tagging for Organization ​
You can add tags to your FeatureGroup for better organization and cost tracking.
ts
const taggedFeatureGroup = await AWS.SageMaker.FeatureGroup("taggedFeatureGroup", {
FeatureGroupName: "TaggedCustomerFeatures",
RecordIdentifierFeatureName: "customerId",
EventTimeFeatureName: "eventTime",
FeatureDefinitions: [
{
FeatureName: "customerId",
FeatureType: "String"
},
{
FeatureName: "eventTime",
FeatureType: "Timestamp"
},
{
FeatureName: "purchaseFrequency",
FeatureType: "Number"
}
],
Tags: [
{
Key: "Project",
Value: "CustomerAnalytics"
},
{
Key: "Environment",
Value: "Production"
}
]
});