一种挖掘不完整数据的新方法
28 浏览量
2021-03-04
11:23:33
上传
评论
收藏 91KB PDF 举报
Vol.30 No.4 JOURNAL OF ELECTRONICS (CHINA) August 2013
NEW METHOD OF MINING INCOMPLETE DATA
1
Wang Lunwen Zhang Xianji Wang Lunwu Zhang Lin
*
(Research Division 404, Electronic Engineering Institute, Hefei 230037, China)
*
(Department of Computer Science and Technology, Anhui University, Hefei 230039, China)
Abstract The data used in the process of knowledge discovery often includes noise and incomplete
information. The boundaries of different classes of these data are blur and unobvious. When these data
are clustered or classified, we often get the coverings instead of the partitions, and it usually makes our
information system insecure. In this paper, optimal partitioning of incomplete data is researched.
Firstly, the relationship of set cover and set partition is discussed, and the distance between set cover
and set partition is defined. Secondly, the optimal partitioning of given cover is researched by the
combing and parting method, acquiring the optimal partition from three different partitions set family
is discussed. Finally, the corresponding optimal algorithm is given. The real wireless signals offten
contain a lot of noise, and there are many errors in boundaries when these data is clustered based on the
tradional method. In our experimant, the proposed method improves correct rate greatly, and the
experimental results demonstrate the method’s validity.
Key words Clustering; Incomplete Information; Partition; Data Mining
CLC index TP18
DOI 10.1007/s11767-013-3006-5
I. Introduction
It is a puzzle to apply data mining on incom-
plete information for it is fuzzy in different mode
edges. What is more, it has intersection in different
mode. However, incomplete information occurs in
every field, so it is of great significance to research
on effective mining method. When clustering on
incomplete information, the mode is always cover
instead of partition. But we hope to receive parti-
tion results in discursion or learning process. Ref. [1]
makes use of transitive closure method to obtain
correspond partition. In this way, although parti-
tion is obtained, it is too coarse to use. Classical
rough set method can not manage incomplete noisy
information effectively. Nowadays, this method has
been extended, such as extended rough set mode
based on tolerance relation, similarity relation, and
limited tolerance relation. However, these extended
1
Manuscript received date: January 10, 2013; revised date:
May 25, 2013.
Supported by the National Natural Science Foundation of
China (No. 61273302) and partially by the Natural Sci-
ence Foundation of Anhui Province (No. 1208085MF98,
1208085MF94).
Communication author:Wang Lunwen, born in 1966, male,
Ph.D., Associate Professor. Research Division 404,
Huangshan Road 460, Hefei 230037, China.
Email: wanglunwen@163.com.
modes have limitations for they only can solve
certain level of problem, they can not manage
complicated incomplete information system
[2]
.
Based on set pair theory, Ref. [3] proposed variable
precision rough set
(),al which provides heuristic
attribute reduction algorithm based on similarity in
positive field. Ref. [4] proposed minimal consis-
tency overriding method. This method fills up
incomplete information and convert overriding
problem to partition problem based on data con-
sistency. But different fill up methods of different
incomplete information will affect reduced number
of rule set. Ref. [5] researched on fuzzy clustering
methods and compared their performance. It pro-
vided Oprimal Completion Strategy Fuzzy C-
Means (OCSFCM) and Nearest Prototype Strategy
FCM (NPSFCM) aimed at incomplete information
and improved clustering results. But the stability is
not good enough. Ref. [6] proposed a visible clus-
tering method aimed at incomplete information to
easy mining data distribution and structure. How-
ever, it must have manual intervention. Ref. [7]
discussed structure and property of fuzzy subset
from granularity and hierarchy concept. These
methods above discussed partition problem of in-
complete information mostly from macroscopic
view, so they are of low efficiency.
资源评论
