Event count and cardinality

With the integration page it is possible to view the total number of events that were sent during the previous day.

The green circles mean that events are being collected while the gray circle means that the particular event type is not being tracked at the moment.

In addition you can also see the number of events per active user, and cardinality for each event type. The visualization will show a cardinality of 0 if no events of that type are being tracked.

info

See more about our event tracking limits here.

What is cardinality

Cardinality in video games can help measure e.g., how many different kinds of actions, items, or events are in the game, and whether it’s a simple or complex system. While high cardinality in game design allows for a rich and complex game world for your players, tracking every single possible combination leads to difficulties in making data driven decisions due to:

• Massive datasets with high dimensionality. This makes it difficult to store, process, and analyze the data.
• Sparse Data. Not all combinations will occur frequently. As a result, you may end up with sparse data for certain combinations, making it difficult to draw reliable conclusions or train predictive models.
• Noise vs. Signal. High cardinality introduces a lot of noise into the dataset. Not every combination of enemy type, combat style, and item type will be meaningful. The challenge lies in identifying the signal (important patterns) from the noise (random or less relevant interactions).
• Feature Explosion. Every additional unique value in a variable (like item types or enemy weaknesses) adds another potential feature in your analysis. With too many features, you might struggle with overfitting or being too specific to certain data points, rather than generalizing trends.

info

If you're playing a game where your character can do different actions, like fight 80 different enemies, with 20 different combat styles, and collect 100 different items. This creates a maximum potential cardinality of 80 x 100 x 20 = 160.000 unique combinations.

To reduce complexity, you can aggregate or simplify your data by:

• Grouping less frequent values into broader categories. For example, grouping enemies by difficulty level or element type instead of treating each type individually.
• Feature Engineering by creating new features from existing ones (e.g., combining enemy type with combat style).
• Feature Selection: Selecting only the most relevant variables from the data.
• Use other event types. GameAnalytics features various event types for tracking in-game interactions. Some events are collected automatically while others need manual integration. For more information refer to our Event types overview.