For the complete documentation index, see llms.txt. This page is also available as Markdown.

Concepts

Technical concept

Computed fields are defined in the schema at the UserPoint level.

The plugin behind it performs computations on a regular basis of a function's result, based on the user activities and profile updates.

Definitions

Computed Field

A computed field is an instance of a Datamart function applied within a specific context. To use it, it must be declared in your schema.

Computed Field Function

A Computed Field Function is a plugin of type COMPUTED_FIELD_FUNCTION, which can be created via the API following the standard process or through the computing console interface.

Notions

A computed field is defined by these elements:

  1. State: The data stored to build the result. It is the history. Each new activity updates the State.

    → the state is closely related to the Lookback Window - i.e. the historical depth of events to be analyzed.

  2. Logic: The logic can be very simple (like a formula, a sum, a count…) or more complex (like a logical operation).

  3. Result: The data calculated and stored. It contains the score(s) to be used.

    The result can be either a single variable or an object, depending on the complexity of the computation. It is computed from the State

Other interesting concepts to take into consideration:

  • Input Data: Either new userActivities, either userProfile update, either both.

  • Period Update: The regularity of updates (every x days - by default x=1).

Lifecycle

  1. INITIAL – The instance is created with this status.

  2. INITIAL_LOADING – Once declared in the schema and successfully validated, an initial loading job is triggered.

  3. ACTIVE – The computed field becomes active if the job is completed successfully.

Drawing
Computed field lifecycle

Technical flow

To better understand how a computed field works, here are the technical steps and the relation with the methods:

  1. Initialization: The computed field is declared in the schema.

  2. State Updates: Triggered by new user activities or profile updates.

  3. Result Calculation: The result is computed based on the current state.

  4. Storage: The result is stored for querying.

Last updated

Was this helpful?