Novel Method of Uncertain Data Modeling and Classification Based on Cloud Model
Qin L · 2014
Data contain inherent uncertainty.Sampling errors,data staling and repeated measurements are all the sources of uncertainty,so the analysis of uncertain data obtains more and more attention in many applications.The value of each traditional uncertain data is represented as domain over which a probability distribution function is defined.Because uncertainty data have fuzziness and randomness,and the traditional probability distribution function is difficult to define the actual distribution of the uncertainty data,this paper proposed a cloud modeling process of uncertainty data by the cloud drops distribution,and also designed a classify method by cloud union and similarity computing of cloud.Cloud model can effectively merge the randomness and fuzziness together,and can analyze uncertain data more effectively.For it's realistic reflection of actual distribution of the uncertain data,our experiments also prove the validity of this method.