Is it possible to use the shortest path to find a route to the cluster center (k-means)?

Hi, I tried to make some modifications in the k-means clustering, using the shortest path(based on weights) to find the center of the cluster, who knows whether it can be done? The diagram below explains what I mean:

Figure 1: use normal kmeans to connect each node to center of cluster

%

Figure 2: use shortest path to connect each node to center of cluster

Please provide advice and resources if any.

Thanks

[Walter Roberson]

No. The center of the cluster is rarely a node in the cluster.

[Asyran Abdullah]

Thanks sir, but is there another way to find shortest path in cluster?

[Image Analyst]

Shortest path from what to where? Are you really wanting to know about solving the traveling salesman problem?

If so, see https://www.mathworks.com/help/optim/examples/travelling-salesman-problem.html?s_tid=srchtitle

Answer 1

[Walter Roberson]

Perhaps you would be interested in https://www.mathworks.com/help/matlab/ref/graph.centrality.html#bu4486d-type Centrality perhaps 'closeness' measure.