Thanks for the email Alf.
You are totally right, there are multiple tools available to perform any machine learning
or data mining analysis such as k-means with in many advanced forms. One of them is weka,
which I used recently and is open-sourced. And just like weka, there are many others.
Let's say I would like to perform a k-mean clustering using native functions mainly for
academic purposes. I know there might be a significant negative impact on performance, at
least in comparison with in-memory like tools.
So, this is why question still remains open. I would like to stick to or discard the
possibility of implementing a k-mean clustering algorithm using native functions.
Thanks again for your reply,
From: Alf. Hester [Alfons.Hester@stripped]
Sent: Thursday, October 17, 2013 12:41 AM
Subject: Re: K-means clustering using native functions
my experiences with K-Means Clustering shows that it is much faster and
more flexible than a (direct) database approach to perform it in
Usually for the most widely used programming languages still solutions
(more or less modular) exist.
Primarily datas path DB -> K-Means Module has to be done.
Am 17.10.2013 07:21, schrieb Nunez Robinson, Claudia Isabel:
> I'm new working with the mysql code and have couple of questions I hope the members
of this mailing list can help me to answer.
> I would like to implement a k-mean clustering function in mysql using the native
functions in mysql. In order to assign an element/tuple to a cluster, I need to do some
pre-computations using all the tuples in a given table. I know that due to the atomicity
property, this is probably very hard to achieve using native functions.
> So, my question is, is it possible to access/read all the tuples in a table from a
native function? If the answer is yes, can you point me to a current function that does
this (I haven't found any so far). If the answer is no, could you recommend another right
way of doing this?
> Thanks for your kind reply,
MySQL Internals Mailing List
For list archives: http://lists.mysql.com/internals
To unsubscribe: http://lists.mysql.com/internals