An actor is the "who or what" that is performing an event. An actor can be a user, physical object (such as a device), digital object (like a subreddit), or a bot.
Actor columns are identified at ingest time by explicitly making them shard keys. When an event source has multiple actor populations, they are implemented through multiple shard keys on the same dataset (table). An actor is one of the “mandatory” fields for every logged event, which also includes a timestamp and name of the event. You can associate properties with an actor that specify the following:
- Aggregate—The most common actor properties aggregate events or flows. Such as, an actor property of count number of sessions for sessions with total time > 5 min.
- Defined Value—If you have several actor properties, you may want to perform a bucketing operation. Such as, if "Num5MinSessions > 10" then "heavy user" otherwise "light user".
- Function—You can perform a mathematical function and apply it as an actor property. For example, if you have Num5MinSessions and Num60MinSessions, you can apply a function that returns a ratio of the two for the actor property.
Defining an actor property
This section demonstrates how to create an actor property. For our example, we create an actor property that determines the movies watched by each user. You can apply the principles from this example to create your own actor property.
To create an actor property, do the following:
- In the left navigation bar, click Actors.
- Choose a dataset from the drop-down list at the top of the window. In this example, we chose the IA_MOVIES dataset. The options available to you will reflect your data.
- Click New Actor Property in the upper right corner of the window.
- At the top of the page, enter a unique name for the actor property. We named our property Movies watched.
- Select an actor from the drop-down list. In this example, we selected username. The options available to you will reflect your data
- Click the Show tab and choose the appropriate options from the drop-down lists. In our example data, we chose count of events.
- Make selections for Filtered to options. We chose events with action that matches watch_movie.
- Specify a trailing window and time range (for more information, see specify relative time in a query). We specified a trailing window of 1 week, and accepted the default time range values of Beginning of time until Now.
- Optional: Specify Defined Value and Function options as necessary. We accepted the defaults for our example.
- Click the Save in the upper right corner of the page, and then click GO. We received the following results for our example actor property.
Our results is rather busy. We could filter over a shorter time range, or other filter options to further explore the behavioral trends in the data.