ℹ ️ Two types of dendrogram

It is important to understand that dendrograms are used in 2 very different conditions that are based on 2 different data inputs:

• Hierarchical dataset: think about a CEO managing team leads managing employees and so on.
• Clustering result: clustering divides a set of individuals in group according to their similarity. Its result can be visualized as a tree, often going together with a heatmap.

Dendrogram for clustering with Matplotlib

The first chart of this section explains how to build a basic dendrogram with Python and matplotlib. It starts from a numeric matrix, compute the similarity between each pair of item thanks to the linkage() function and plot the result with the dendrogram() function.

Most basic dendrogram and clustering with Python and Matplotlib

The next post explains how to customize the output: change node and edge features like color and shapes, select a specific part of the tree and more.

Most basic dendrogram and clustering with Python and Matplotlib

Horizontal dendrogram

Cut the dendrogram at a specific height

Dendrogram, clustering and heatmaps with Seaborn

It is very common to display the results of a clustering using a heatmap with dendrograms around it. Fortunately the Seaborn library makes it a breeze.

The examples below will guide you through the most usual needs. Note however that data normalization should be your main focus here, as it can massively impact the final visual result. (See this post for that).

A seaborn heatmap with clusterization and dendrogram applied

Most basic heatmap with dendrogram with seaborn

Change color palette

Learn how to normalize data in the heatmap, avoiding that kind of result

Dendrogram for hierarchical datasets

Dendrogram can also be used to display a hierarchy. A hierarchy as a root node that gets divided in several children and so on.

TODO