[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: Run kmeans for different random starting values

From   Serguei Kaniovski <[email protected]>
To   [email protected]
Subject   st: Run kmeans for different random starting values
Date   Wed, 9 Jan 2008 13:56:13 +0100

Hello All,

the start()-option in the kmeans allows specifying how the initial values 
for the centroids are generated. I am using the random option. As I 
understand, when the command is called the initial values are generated 
only once.

What is the simplest way to repeat the command say 1000 times for 
different random initial values, and store the clusterings? More 
importantly, how do you decide which clustering should be preferred? 
Should I choose the one that occurs most frequently? Also, how to detect 
repetitions which are identical up to the labels (numbers) of the 

Thanks you for you help,

*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index