Introduction

Imagine a diverse group of individuals engaging in a physical activity intervention. You collect data on their daily steps, their active minutes per day, and their calories burned per day.

By conducting a cluster analysis, you could ‘group’ the individuals into three clusters or ‘types’, for example Cluster One (red - higher calories burned with a range of daily steps), Cluster Two (green - higher daily steps but moderate calories burned) and Cluster Three (blue - lower daily steps and lower calories burned).

This is really useful, because it helps us start to identify different ‘types’ of responses on multiple variables, or different groups within our data. This might in turn lead to some further research questions or analysis on those clusters.

<aside> 💡

Therefore, cluster analysis identifies patterns in our data by organising it into subsets that share common characteristics.

</aside>

Content map

To learn more about cluster analysis, read each of the following sections:

Partitioning vs. Hierarchical models

Density-Based vs. Distribution-Based Clustering

Hard vs. Soft Clustering

Feature Selection and Dimensionality Reduction

Next element

Measuring similarity