Understanding Features and Entities

In Predicti Data, a feature is a fundamental data model representing a single measurable or descriptive value tracked over time. Features consist of historical data points, capturing how a specific attribute changes at different timestamps.

---

What Is a Feature?

A feature consits of one or more values, each associated with a timestamp forming a time series of the attribute's state or measurement.

Feature Structure Example

{
  "history": [
    { "2006-03-08T14:26:56Z": ["2"] },
    { "2007-05-01T10:15:00Z": ["3"] }
  ],
  "id": "11f17ae3-be00-0668-e044-0003ba298018",
  "_lastModified": "2025-06-03T11:21:46.879Z"
}

• history records changes over time: each entry links a timestamp to a list of values (usually a single value per timestamp).
• id uniquely identifies this feature instance.
• _lastModified tracks the last update time of the feature.

This structure enables precise tracking of attribute changes, supporting longitudinal analysis and up-to-date status retrieval.

What Is an Entity?

An entity is a logical grouping or container for related features that collectively describe a real-world object or concept, such as:

• An address
• A person
• A property
• A household

Entities organize features into meaningful sets, making it easier to navigate, query, and manage data.

Example: dk-address Entity Features

• address_houseNumber
• address_postalCode
• address_city
• address_floor
• address_door

Each of these features independently tracks its values over time, maintaining a historical record.

See the DK Address Data Features documentation for more details on the address entity and its features.

Why Use Features with History?

Tracking features as time series allows you to:

• Understand how values evolve (e.g., changes in residency type, property size, or postal codes).
• Perform trend analysis and detect patterns over time.
• Access the most current state of an attribute as well as its history for auditing and verification.
• Build dynamic and temporal models that react to real-world changes.

Summary

Concept	Description
Feature	A single attribute that consists of values with timestamps (time series)
Entity	A collection of related features describing a real-world object

Together, entities and features provide a powerful framework for managing rich, time-aware data that supports accurate and flexible insights.