X-Means producing different clustering with each run

[rtrad89]

I am sorry if the issue is redundant; I have rifled the issues for a similar concern but alas, to no avail.

Is there a way to pause the randomness in X-Means, so that the clusterings it produces are consistent (perhaps through fixing the random_seed or random_state)? Because I am doing some experiments and it's elusive if the X-Means results keep varying run after run.

Many thanks in advance!

[annoviko]

Hello, @rtrad89 ,

Currently, there is no such option. When X-Means tries to split clusters it should chose new centers for them and for that purpose K-Means++ is used.