| Summary: | 碩士 === 世新大學 === 資訊管理學研究所(含碩專班) === 102 === Statistical data has been used in several concepts,like medical、health、public、manufacture,and the statistical analysis report also is important to decision.
Over the past few years, data mining has often used in analysis statistical data, and one of the most popular is clustrering.The traditional clustrering method used all the variables to clustrer,but the result of clustrering would be influenced by the variables, because the number of variables are not equal,it means the weight of attribute might not be correct.
We propose a new method to Clustering of statistical data, two stage of SOM(Self Organizing Map).In first stage,we reduce the variables from several to single depend on the similar of data. In second stage, we use the result of first stage to cluster.The new method can improve the efficiency of clustering in statistical data.
|
|---|
