64.txt

This complete matlab for neural network

发信人: NAOMIELIE (雁来红), 信区: DataMining
标  题: Re: 请问分类和聚类的一个问题
发信站: 南京大学小百合站 (Sat Mar 29 00:05:40 2003)

I think there are quite a lot of clustering algorithms are
designed for categorical data.

One interested one is written by Kleinberg et al 
(he is not the first author, maybe), named as 
categorical data clustering: a dynamic system approach.

And, Joshua Huang used to invent two algorithms,
named as K-modes and K-prototypes, which are designed for categorical
data and mixed data (data with both numerical and categorical data)

Wiley & Sons Publisher published a series of books on statistics,
in which there are two by the same person.
An Introduction to Categorical Data Analysis
and
Categorical Data Analysis.

The second one is in more details. 
The author thinks it is too hard for most readers,
so just prsent some main results in the first one.
And he add some new parts in the first book.
You can learn how statistics people deal with categorical data.

Other related things can be found by using keywords:
'metric space'+clustering or
hamming distance,
etc.
However, you may not obtain what you want in the search result,
although some valuable things may appear.

goodluck.

【 在 odd 的大作中提到: 】
: 怎??理符号?量。
: 就是说针?向量 X(x1,x2,.....,xn)进行分类，里面的分量
: x1到xn大部分是符号分量，怎?计算向量间的距离?？
:    一个例子就是客户的属性向量，地址?，爱好?，等等。怎?计算
: 客户间的距离？

--

※ 来源:．南京大学小百合站 http://bbs.nju.edu.cn [FROM: 210.172.103.218]