Method of indicator reduction based on cluster analysis and
rough set theory and its application
BAO Xin-zhong
(Dongling School of Economics and Management, University of Science and Technology
Beijing, Beijing 100083, China)
Abstract: This paper introduced a new method of indicators reduction to avoid information redundancy and improve research efficiency. Firstly, a method of eliminating indicators’ correlation based on cluster analysis was introduced. Secondly, rough set was used to choose indicators among the same category according to their significance. A final system of indicators came into being afterwards. In traditional researches on indicators reduction, either correlation or significance of indicators was taken into account. The method takes not only correlation but also significance of indicators into consideration and thus it is more reasonable. To test this new method, an example of selecting a system of indicators for financial distress forecasting was given after the introduction of this new method. 32 indicators were reduced to 9, showing that this new method was feasible and effective.
Key words: Indicators reduction; Cluster analysis; Rough set; Correlation; Significance
1 Introduction
Indicator system can have much impact on results of empirical research. From the aspect of quality, the more relative the indicators to research, the bigger the explanatory power of models will be. From the aspect of quantity, few indicators can lead to small explanatory power while too many may provide redundant information and reduce model effectiveness. Besides, correlation between indicators can affect model efficiency. So it is very important to select indicators of importance that can also guarantee model’s explanatory Power by reasonable methods. There are typically three ways to establish an indicator system: (1) reducing by subjective experience which depends much on researchers’ experience; (2) reducing by indicators’ importance which means to choose indicators that have influences on decision-making; and (3) reducing by indicators’ correlation which means to choose one indicator among those highly correlated. We focus on the second and the third way as the first one is too subject.
Reducing by indicators’ importance is the mainstream, and rough set theory is widely used in this field.
Richard Jensen, Qiang Shen (2004) present a novel approach based on an integrated use of fuzzy and rough set theories to reduce data redundancy and apply it to web categorization. Experimental results show that fuzzy–rough reduction is more powerful than the conventional rough set-based approach.
Chen Degang et al (2007) put forward consistent and inconsistent covering decision systems and their attribute reductions. They finally use a discernibility matrix to design algorithms that compute all the reductions of consistent and inconsistent covering decision systems. Better classification performance of the proposed attribute reductions of covering decision systems than those of traditional rough sets is proved by numerical tests.
Zengtai Gong et al (2008) deal with an interval-valued fuzzy information system by means of integrating the classical Pawlak rough set theory with the interval-valued fuzzy set theory. They
rough set attribute reduction method based on the... the utilization of cluster analysis could ...Key words: rough set theory; attribute reduction;...
A new method for rough set-neural network based...based on reduction reduces the dimension of input...(cluster analysis)等方法 的评价结果仅对方案决策或...
Based on Variable Precision Rough Set Lu Lin ...Analysis and Processing Method of A Data Stream ...Reduction Lei Huang, Changping Liu Multi-camera ...
describedmethodofsoft veryflexible.Inthispaper,...sharingtoclusteranalysis,By transformed ...soRsettheoryandparametersreductionbased on relational ...
Analysis and Visualization of Large Networks using...(Critical Path Method); extracting subnetwork; ... reduction: hierarchy, subdivision, degree; ...
We present a method for simultaneous dimension reduction and metastability analysis of high dimensional time series. The approach is based on the combination ...
rough sets, cluster analysis, genetics algorithms...The theory is based on sound mathematical ...Further, a very important issue, reduction of ...
concept of attribute reduction in the rough set ..., which is based on statistic learning theory. ...Cluster Analysis and Support Vector Machines 狄明明...
indicator of economic development since more than ... and adopts cluster analysis to determine whether ...SFA is based on the quantitative economy theory ...
With the proposed method, the elements of each ...Cluster analysis is one of the model reduction ...Then, spectral theory is applied to identify the...
热门文档