發(fā)信人: GzLi (笑梨), 信區(qū): DataMining
標  題: [合集]學習數(shù)據(jù)挖掘之前要修什么課？？
發(fā)信站: 南京大學小百合站 (Wed Sep 11 12:42:44 2002), 站內(nèi)信件

softgenius (我愛linux) 于Wed Sep  4 15:52:20 2002)
提到：

我是外系的學生，不懂的問。

一般考試是如何考阿？？？

strawman (獨上江樓思渺然) 于Wed Sep  4 16:03:49 2002提到：

知道一些數(shù)據(jù)庫的知識，還有數(shù)理統(tǒng)計的知識，我想就可以了吧。
當然也要看教材的深淺了。


sinokdd (KDD in China) 于Thu Sep  5 12:38:37 2002)
提到：


I donot think you need to know Database, actually the last several

chapters in Prof. Han's book can be taken as the introduction of

data mining techniques, and after that you can read 

Machine Learning by Tom Mitchell, then you can read some papers for

the topic that you are interested in, for example, C45, clsutering,

etc.



jimo (寂寞) 于Thu Sep  5 17:40:50 2002提到：

關聯(lián)規(guī)則就不需要太多機器學習的東西



sinokdd (KDD in China) 于Fri Sep  6 08:14:45 2002)
提到：


Yeah, but data mining is not association rule or sequential pattern only.

If you only know AR or SP, I am sure there is very little thing

you can do for data mining.

And actually data mining is borrowed from statistics, and of course you can

say that AR is nothing to do with statistics.

The definition of KDD is the process of pattenr discovery from large data,

the mostly used techiques are from machine learning and statistics. And now

people donot care what technology you use in mining, but if you can get some


patterns, not the original data itself, then you can call the process as

data mining.



jimo (寂寞) 于Fri Sep  6 12:09:19 2002提到：

呵呵   ar 里的sampling 和統(tǒng)計還是有關系的



helloboy (hello) 于Sat Sep  7 17:01:15 2002提到：

Sampling is a process which take a small part of data from origin dataset.
Statistic is different.It use maths model to evaluate data.



fervvac (高遠) 于Sun Sep  8 00:36:06 2002提到：

there are theories about "sampling", which was built on top of statistics
stuffs.



helloboy (hello) 于Sun Sep  8 08:24:04 2002提到：

So sampling is based on statistics. right?


fervvac (高遠) 于Sun Sep  8 18:28:40 2002提到：

Yes, you can say so.

In fact, if you look at the use of sampling in database literatures, much of
the paper is used to derive the (probabilistic) bound.