Using two differents distance in the same clustering algorithm

Hello,

I'm trying to create a clustering algorithm using a particular approach:

First, I want to use the mahalanobis distance to compute distance between observations (I do have the ML and optimization toolbox).

Then, my idea is to use the linkage function to create a tree. But the issue that I cannot find an answer for is that I want to compute distance between nodes using a distance that I created, which is different from the mahalanobis distance. The second distance is only working between clusters.

Thanks for your answers

Answer 1

[Dheeraj Singh]

I understand that you want to use mahalanobis distance for calculating distance between clusters and then use a custom distance function to create a tree using linkage function.

Firstly, you can use pdist function to calculate distance between clusters using mahalanobis distance.

While creating a tree using linkagefunction you can define the custom distance function as a function handle. For more info about the different distance function please refer to the Distance metric section of the documentation of linkagefunction .