Skip to main content

Attribute groups

Define the behavior you want to capture in attribute groups.

Attribute groups by data source​

Each of the three available data sources has its own attribute group class. Choose which one to use depending on how you want to calculate and sync the attributes.

Use a StreamAttributeGroup to calculate attributes from the real-time event stream.

from snowplow_signals import StreamAttributeGroup, domain_sessionid

my_stream_attribute_group = StreamAttributeGroup(
    name="my_stream_attribute_group",
    version=1,
    attribute_key=domain_sessionid,
    owner="user@company.com",
    attributes=[
        # Previously defined attributes
        page_view_count,
        products_added_to_cart_feature,
    ],
)

Attribute group options​

The tables below list all available arguments for each type of attribute group:

ArgumentDescriptionTypeDefaultRequired?
nameThe name of the attribute groupstring✅
versionThe version of the attribute groupint1❌
attribute_keyThe attribute key associated with the attribute groupAttributeKey✅
ownerThe owner of the attribute groupEmail✅
descriptionA description of the attribute groupstring❌
ttlTime-to-live for attributes in the Profile Storetimedelta❌
attributesList of attributes to calculatelist of Attribute✅
onlineCalculate attributes (True) or not (False)boolTrue❌

If no ttl is set, the attribute key's ttl will be used. If the attribute key also has no ttl, there will be no time limit for attributes. We highly recommend setting a TTL. Our suggested default is 7 days for stream attribute groups, and 365 days for batch attribute groups.

Use the online property to control whether or not Signals should actively compute the attributes, or just register the configuration.

Versioning​

Use version=1 for the first version of an attribute group. After publishing, if you want to change the definition in any way, iterate the version number.

Testing attribute groups​

To understand what the output of an attribute group will look like, use the Signals test method. This will output a table of attributes calculated from your atomic events table.

from snowplow_signals import Signals

# Connect to Signals
sp_signals = Signals(
        {{ config }}
    )

# Run the test
test_data = sp_signals.test(
    attribute_group=my_attribute_group,
    app_ids=["website"] # The app_id in your Snowplow events
)
note

While you can filter on specific app_ids during testing, both the streaming and batch engines may be configured to process only a subset of relevant app_ids to avoid unnecessary compute. As a result, testing with an arbitrary app_id may not yield expected data if it isn’t included in the configured subset.

To see which attributes an attribute group has, use get_attribute_group(). Here's an example:

attribute_definitions = sp_signals.get_attribute_group(
    name="my_attribute_group",
    version=1,
)

print(attribute_definitions)

The table below lists all available arguments for get_attribute_group()

ArgumentDescriptionTypeRequired?
nameThe name of the attribute groupstring✅
versionThe attribute group versionint❌

If you don't specify a version, Signals will retrieve the latest version.

Publishing attribute groups​

Use the publish() method to register attribute groups with Signals. This makes them available for real-time calculation and retrieval.

from snowplow_signals import Signals

# Connect to Signals
sp_signals = Signals(
        {{ config }}
    )

# Publish attribute groups
sp_signals.publish([
        my_attribute_group,
        my_other_attribute_group
    ])

If you only want to publish the attribute group definitions, and don't want to calculate the attribute values now, set the online=False option.

Attribute groups can be set to expire​

Some attributes will only be relevant for a certain amount of time, and eventually stop being updated.

To avoid stale attributes staying in your Profiles Store forever, you can configure TTL lifetimes for attribute keys and attribute groups. When none of the attributes for an attribute key or attribute group have been updated for the defined lifespan, the attribute key or attribute group expires. Any attribute values for this attribute key or attribute group will be deleted: fetching them will return None values.

If Signals then processes a new event that calculates the attribute again, or syncs the attribute from the warehouse again, the expiration timer is reset.

Extended stream attribute group example​

This example shows all the available configuration options for a stream attribute group.

This attribute group groups attributes for a user attribute key, to be calculated in real-time.

from snowplow_signals import StreamAttributeGroup, user_id

stream_attribute_group = StreamAttributeGroup(
    name="comprehensive_stream_attribute_group",
    version=2,
    attribute_key=user_id,
    owner="data-team@company.com",
    attributes=[
        page_view_count,
        session_duration,
        conversion_rate,
    ],
    description="User engagement attributes in real-time",
    ttl=timedelta(days=90),  # Attributes live in the Profiles Store for 90 days
)

Signals will start calculating attributes as soon as this attribute group configuration is applied.

On this page

Want to see a custom demo?

Our technical experts are here to help.

Book a demo