Cluster analysis groups similar objects together into clusters. It’s useful in many different fields: for identifying customer segments or even for studying evolution.
There are 2 main types of clustering: agglomerative and divisive.
- Agglomerative clustering gradually merges similar items together into bigger and bigger clusters
- Divisive clustering takes a large cluster and gradually divides it into smaller and smaller clusters
First, a statistical method is chosen to identify the differences between objects. Euclidian distance is the most common, which is the Square root(SUM(first point’s distance to origin – second point’s distance to origin)^2) Distance is nothing more than the difference between two clusters. In the marketing data example (if we’re looking at customer ages), distance would be the difference in ages.