Factor interval data analysis and its application
Huiwen Wang, Henry M.K. Mok, Dapeng Li · 2007
In symbolic data analysis, interval data is a very important type of data, which can extract the tendency of centralization and dispersion of a dataset. When applying multivariate analysis on an interval dataset with n × p dimensions, the dataset can be described as a hyperrectangle with n × 2 p vertices. In some circumstance of application, the calculation work will rocket up as the increasing value of dimension p . Additionally, the hyperrectangle may enlarge the range of the original dataset and reduce analysis accuracy. To better resolve these problems, factor interval data, a new type of symbolic data, is studied in this paper. First, the definition and extraction approach of factor interval data is presented and its advantages are discussed. Then, we apply this new symbolic data and its PCA method on Chinese stock markets. The results have showed that, by using the method of this new symbolic data, the research results are very consistent with the realistic characteristics of Chinese stock markets, which proves that factor interval data is a very easy and effectual way to simplify the multidimensional data system.