What is Cluster Analysis?

What is grouping?

Cluster analysis fundamentally refers to the process of grouping sets of objects in such a way that objects assigned to the same group, called a cluster, are more similar (in some sense or another) to each other than to those in other clusters. In game data science, the objects can be players, agents, quests, etc. Each of these is described via a finite set of variables.

Example

For example, for a player, there can be many variables, such as the number of kills or deaths, their character level, their XP, and movement patterns. As discussed in the earlier chapter, we often call these variables features.

What is a cluster?

The term “cluster analysis” is a process describing an algorithm used to develop clusters. There are many algorithms developed for this task, each with different strengths and weaknesses and with different traditional application areas in different fields. Therefore, the definition of what a cluster is also varied between algorithms. While all of these algorithms somehow refer to a group of objects in a data space, understanding the different cluster models is vital to be able to apply the algorithm to the data correctly.